ACSSA Aware No. 25, 2010


Measuring sexual offender recidivism

By Mary Stathopoulos

A key dimension of sexual assault prevention is stopping perpetrators from re-offending (often referred to as tertiary prevention). Treatment programs that aim to stop re-offending are available for adult and juvenile sex offenders throughout Australia (MacGregor, 2008). There are also numerous monitoring and surveillance mechanisms aimed at released offenders in various states and territories, the purpose of which is to minimise the chances of re-offending.

One way of monitoring effectiveness in this regard has been through the use of recidivism rates, defined below. Rates of re-offence also assist in understanding and addressing issues in the criminal justice system (Australian Bureau of Statistics [ABS], 2001). Yet, results from international studies vary quite dramatically, with recidivism rates of between 3% and 70% being reported (Lievore, 2004). In Australia, sexual recidivism rates range from "two percent in some samples to as high as sixteen percent in others" (Lievore, 2005, p. 2). Overall, this is considered low in comparison to other types of offences.1

What are the reasons for this variation and what can we reasonably consider recidivism rates to tell us? This summary describes how sexual offender recidivism may be measured and discusses the impact this may have in how rates are used.


In general, recidivism refers to a relapse or return to criminal behaviour. However, for the purposes of measuring recidivism, this definition varies considerably. It varies first in terms of the point in the criminal justice process at which it is measured. Thus, "recidivism" may be operationalised as:

  • re-offence;
  • re-arrest;
  • reconviction; and/or
  • return to prison.

Although measuring recidivism on the basis of re-offence would provide the most accurate measure, this is extremely difficult to ascertain if it is not brought to police attention and if it is not recorded as an offence. Measuring recidivism on the basis of reconviction happens much later in the justice timeline - arrest, charge, prosecution, trial/plea and conviction need to have occurred before it is counted. Re-entry into prison is farther yet again and relies on sentencing decisions. The latter two measurement criteria are more conservative than arrest data and are unlikely to correlate with actual re-offending (Lievore, 2004)

A second variation in measurement is the repeat offence that is being counted. For sexual offenders, this could be:

  • the same type of offence as the original (e.g., rape);
  • another sexual offence generally (e.g., child sex offences, indecent assault, rape);
  • another violent offence (e.g., assault causing serious injury); or
  • any other offence (e.g., motor vehicle theft).

Decisions about how to operationalise recidivism may be informed by the purpose of the research or analysis. Looking at re-conviction for any other offence, such as violent crime, can be useful in understanding juvenile offenders. Lievore (2005) stated that juvenile offenders are at risk of "growing into" offending behaviours. Identifying and treating adolescents who have visibly offended can help break a cycle that may continue over a lifetime. They may also be informed by the accuracy of the data itself. For example, re-offending that is counted at re-arrest or warrant will yield higher recidivism rates, but the information is patchier because of jurisdictional differences in recording, and missing data. On the other hand, reconviction measures provide a more complete record, but are underestimations.


Researchers use a variety of methods/approaches to arrive at a rate of recidivism.

A comparative analysis between treated and untreated offenders may offer an insight into how to prevent recidivism through perpetrator treatment programs. This can be approached in different ways. A randomised control trial may be used, however the ethical implications of denying offenders who wish to be treated in order to establish a control group are questionable (Marshall & Marshall, 2007). A way to avoid this is to use an incidental untreated sample with which to compare against a sample of sexual offenders in a given treatment program. An incidental untreated sample is a group of incarcerated sex offenders who happen to not have attended any offender programs with which the treated group can be compared. However, issues of equivalence between the groups in relation to static and dynamic risk factors can affect the validity of the research.

Static risk factors refer to fixed variables such as the offenders' sex, age, ethnicity, criminal history and their relationship to the victim. Dynamic risk factors refer to those variables open to change through treatment, including factors such as "substance abuse, general social skills, sexual arousal patterns and the quality of relationships" (Lievore, 2005, p. 2). In order to avoid these issues, another methodology adopted is one based on actuarial risk. Actuarial risk assessment measures come from an evaluation of dynamic and static risk factors. These can then be used to calculate an estimated actuarial risk rate with which to measure the effectiveness of treatment programs or to compare to untreated sexual offender recidivist rates (Marshall & Marshall, 2007).

Regardless of which methodological approach is used, accurate measures are affected by other factors, such as follow-up periods, drop-out rates, alternative and hidden offences and plea bargaining, as well as the data sources used:

  • Follow-up periods
    Follow-up periods refer to the length of time that sexual offenders are "trailed" in terms of their sexual offending behaviour. Follow-up periods can be anywhere from six months to twelve years. The longer the follow-up period allowed, the more accurate and valid will be the measure of recidivism. Follow-up periods can be affected by whether the research is retrospective and the time and economic constraints to the researchers. Retrospective as well as prospective studies of sexual offending recidivism can be plagued by low response rates and drop-out rates of sexual offenders.
  • Drop-out rates
    Drop-out rates may be due to selective attrition or geographical issues (Hanson, Broom, & Stephenson, 2004). These concerns also contribute to the difficulty of measuring treatment efficacy.
  • Alternative and hidden offences
    Alternative and hidden offences can also affect outcomes. If a sexual offender is re-arrested or re-convicted for an alternative sexual offence or another type of offence, rates of recidivism may be affected. As stated above, this can depend on the definition that is given to recidivism at the beginning of the research. Hidden offences are those that are not reported. Lievore (2005) has pointed out that within relationships of previous sexual offenders, coercion and violence in sexual activities may be viewed or experienced as normative and therefore never come to the attention of authorities.
  • Plea-bargaining
    Plea-bargaining refers to an agreement by the offender to plead guilty to a lesser offence in order for the prosecutor to secure a conviction without a trial. Plea-bargaining of offences can also contribute to the hidden nature of sexual offences. A re-offender who is arrested and convicted may plea-bargain to a lesser offence or only be charged for a more violent offence in a multiple-offence scenario.
  • Data sources
    Data sources can include official and unofficial records of sexually offensive behaviour. Unfortunately, there may not be consistency in reporting methods from region to region or across states and territories. Some records may be incomplete and others can sometimes be lost, leaving large gaps in the data required for calculation of sexual offence recidivism (Furby, Blackshaw, & Weinrott, 1998).

Attrition of sexual offences from the legal system

The dropout of sexual assault cases from the criminal justice system—particularly in the early stages—needs to be factored in when interpreting the lower recidivism rates.

The following process of drop-out before arrest or charge affects, from the outset, the size of the population upon which recidivism rates are based. Briefly:

  • 1 in 6 women who experience a sexual assault report to police;
  • two-thirds of reported cases are actually recorded by police (calculation based on Gelb, 2007); and
  • for incidents of sexual assault that were recorded, the offender(s) were proceeded against for approximately 1 in 4 victims (measured at 6 months after the report was made; ABS, 2004).

In addition, it is unclear whether the alleged offenders against whom the police proceeded are representative of sexual offenders who do not come to police attention. Cases that do proceed are more likely to involve: additional physical injury, an unknown perpetrator, a prior offending history of the offender, and forensic evidence (Heenan & Murray, 2007; Lievore, 2005). This does not represent the empirical picture of sexual assault provided by the Personal Safety Survey (ABS, 2006).

Figure 1 shows the decreasing size of the offender population at different stages of the justice process.

What can sexual offender recidivism rates tell us?

Rates of sexual offender recidivism are unlikely to be the whole picture in terms of re-offending. It is difficult to say whether recidivism rates are about sexual offending per se, or whether they only tell us something about the repeat offending of those most likely to be in the justice system in the first place. This is primarily due to how sexual offending comes to the attention of the legal system. That only a sixth of known sexual assaults are reported (ABS, 2006) means that detected offenders are the minority. Where recidivism is defined as reconviction, we can see from Figure 1 how small that population is relative to the number of known sexual assault victims (0.9%). Recidivism rates also cannot tell us about hidden sexual assault such as intimate partner rape. This type of offence, which may be repeated over years, may never come to the attention of police or end up in the justice system. Therefore, the true extent of these crimes and recidivist rates are not currently known.

Figure 1: Attrition of sexual assault cases from the criminal justice system

Figure 1 - Attrition of sexual assault cases from the criminal justice system

Source: Gelb (2008), p.4

Sexual offence recidivism rates can, however, tell us about visible offenders and the points at which they come into contact with the criminal justice system. Although statistics highlighting the rate of recidivist activity of sexual offenders are affected by variables, as is evidenced in Table 1, they can be used to understand the efficacy of treatment programs for offenders. Strategies to assist in increasing the reliability of recidivism rates of sexual offenders include researchers drawing on a range of unofficial data sources, such as the self-reported data of offenders. Self reported data can help fill in the gaps when other official records are not available. Longer follow-up periods such as 20 or 30 years can be established, as longer observation times afford greater periods in which to observe and record criminal activity (Payne, 2007). Finally, adopting consistent recording procedures nationwide so that data can be aggregated across states and territories ensures that all researchers are working with consistent data, and any changes in rates can be captured.

Table 1: Recidivism rates across a sample of follow-up studies



Sample (n)

Index sex offence

Definition of recidivism

Follow-up period

Recidivism offence type

Sex offences

Violent offences

Sex and/or violent offences

Any offence

Burgoyne 1979



Rape (62%)

Attempted rape


49 years

Rape/attempted rape 2%

Other 5%

31% (may have included sex offences)



Broadhurst & Maller 1992



Rape, carnal knowledge, incest, indecent dealings


Up to 12 years

Homologous 5.2%

Other 3.2%




Broadhurst & Loh 1997



Not specified


Average 6 years, maximum 11 years





Greenberg, Da Silva & Loh 2002


2,165 referred for sex offender treatment

Child molesters and rapists


Up to 7 years

Overall 15.5%

Treated rapists 7%

Untreated rapists 4.5%

Treated child molesters 16%

Untreated child molesters 5.6%


Overall 28.3%

Treated rapists 21.2%

Untreated rapists 15.7%

Treated child molesters 18.6%

Untreated child molesters 10.5%

Overall 49.7%

Treated rapists 60.8%

Untreated rapists 60.1%

Treated child molesters 43.3%

Untreated child molesters 41%

Southey, Braybrook & Spier 1994

New Zealand

273 in two groups

Not specified


5 years and 10 years

Rape within five years 6%

Various other sexual offences 4%–13%

33% of each group within five years


Overall 72% over 10 years

Within 5 years: 59% of 10-year group

64% of 5-year group

Spier 2002

New Zealand


Not specified


Up to 5 years

3% within 2 years

7% within 5 years

4% within 5 years for serious violations

12% within 2 years

23% within 5 years


30% within 2 years

47% within 5 years

Motiuk & Brown 1996



Federally sentenced rapists (n = 118)

Paedophiles (n = 114)

Incest offenders (n = 46)


Average 3.5 years

52 months for 329 under community supervision

1–3 years for 241 newly released

Overall 5%

Under community supervision

Rapists 6%

Paedophiles 10%

Incest 4%

New release

Rapists 8%

Paedophiles 4%

Incest 3%

Overall 20%

Under community supervision

Rapists 21%

Paedophiles 18%

Incest 9%

New release

Rapists 25%

Paedophiles 11%

Incest 11%


Overall 34%

Proulx et al. 1998



70 rapists & 102 child molesters taking part in 2-year treatment program (Dropouts <12 months; Completed treatment 12–24 months; Extended treatment >24 months)


1 month to 155 months

Mean of 56 months

Child molesters

Dropout 21%

Completed 6%

Extended 21%


Dropout 21%

Completed 35%

Extended 23%

Child molesters

Dropout 21%

Completed 6%

Extended 27%


Dropout 54%

Completed 50%

Extended 27%


Child molesters

Dropout 33%

Completed 6%

Extended 32%


Dropout 71%

Completed 65%

Extended 39%

Soothill & colleagues 1998, 1999, 2000

England & Wales


Cohort of male sex offenders convicted in 1973


32 years 1963 to 1994


For any sexual offence 17% within 20 years

For serious sexual offences 10% within 20 years



38% within 20 years


61% within 20 years

Marshall 1997a

England & Wales

Men born in 1953

Not specified


Calculated in relation to estimated sex offender population

10% within five years of first conviction

12% within five years of first conviction

22% within five years of first conviction


*A variety of measures was used. Only selected results are reported here.


What recidivism rates can tell us depends on how the definition of this is operationalised. Notably, they cannot tell us about the hidden cases of sexual assault—that is, those do not come into contact with the justice system—and there is a question about whether rates in fact tell us about re-offending per se. They can, however, map the contact that an offender has with the justice system, providing information about points of contact and, together with other research, the surrounding circumstances in offenders' lives that may have led to this re-contact (e.g., Maruna, 2001).


Australian Bureau of Statistics. (2001). The national crime justice statistical framework, 2001 (Cat. No. 4525.0). Canberra: ABS.

Australian Bureau of Statistics. (2004). Sexual assault in Australia: A statistical overview. Canberra: ABS.

Australian Bureau of Statistics. (2006). Personal Safety Survey 2005. Canberra: ABS.

Furby, L., Blackshaw, L., & Weinrott, M.R. (1998). Sex offender recidivism: A review. Psychological Bulletin, 105(1), 3–30.

Gelb, K. (2007). Recidivism of sex offenders research paper. Melbourne: Sentencing Advisory Council.

Hanson, R. K., Broom, I., & Stephenson, M. (2004). Evaluating sex offender treatment programs: A 12-year follow-up of 724 offenders. Canadian Journal of Behavioral Science, 36(2), 87–96.

Heenan, M., & Murray, S. (2006). Study of reported rapes in Victoria 2000–2003: Summary research report. Melbourne: Office of Women's Policy, Department for Victorian Communities.

Lievore, D. (2004). Recidivism of sexual offenders: Rates, risk factors and treatment efficacy. Canberra: Australian Institute of Criminology.

Lievore, D. (2005). Thoughts on recidivism and rehabilitation of rapists. University of New South Wales Law Journal, Forum: Sexual Assault and the Law, 11(1): 293–298.

Marshall, W. L., & Marshall, L. E. (2007). The utility of the random controlled trial for evaluating sexual offender treatment: The gold standard or an inappropriate strategy? Sex Abuse, 19, 175–191.

Maruna, S. (2001). Making good: How ex-convicts reform and rebuild their lives. Washington, DC: American Psychological Association Books.

Macgregor, S. (2008). Sex offender treatment programs: Effectiveness of prison and community based programs in Australia and New Zealand (Indigenous Justice Clearinghouse Research Brief). Sydney: NSW Attorney-General's Department.

Payne, J. (2007). Recidivism in Australia: Findings and future research. Canberra: Australian Institute of Criminology.

Mary Stathopoulos is a Research Officer at the Australian Centre for the Study of Sexual Assault.


1 There is some debate about whether this signifies effectiveness in treatment programs and law enforcement measures or whether it reflects the difficulty of detecting sexual offences and/or redoubled efforts by offenders to conceal their offending.