FORUM on Corrections Research

When Are Sex Offenders at Risk for Reoffending? Results of Two Long-Term Follow-up Studies

Two recent studies found that sex offenders may reoffend for many years after release. In an American study that followed up a group of sex offenders for four years, the critical year for problems reoccurring was the third year after release. In a Canadian study that followed up child molesters, the greatest risk period was the first 5 to 10 years after release.

This research underlines the importance of long-term analyses in evaluating treatment outcome with sex offenders.

American Study

This study is an extension of a previous examination of the effectiveness of an out-patient treatment program that followed up a group of sex offenders for 6 months to 10 years. Using self-report questionnaires and data from the Bureau of Criminal Apprehension, that study found a recidivism rate of only 3.7% among program participants.

The present study uses some of the same information, but also conducted annual interviews with the sex offenders, from six months to four years after treatment. The study looked at the extent of therapeutic change on 10 variables and at the trend in these changes to see which year after treatment, if any, was the most critical. It also looked at the sex offenders' use of a maintenance (relapse prevention) plan and their ability to recognize the early warning signs for sexual reoffending.

Method

There were 70 subjects in the study, mostly pedophiles but also incest offenders and exhibitionists. All had completed a long-term out-patient treatment program, with the length of time in the program averaging three years.

The 70 men agreed to be tested before and after treatment. The number who were interviewed and tested after treatment varied, from 65 in the first year to 28 and 29 in years three and four. Fewer men were available in the later years because some had not been out of the program long enough. Of the original 70, 15 dropped out of the study.

On average, each sex offender was interviewed three times. In all, 214 interview questionnaires were completed. The interview covered offending behaviour, use of a maintenance or prevention plan, psychological factors related to offending and interpersonal factors regarding relationships and work interactions.⁽¹⁾

Results⁽²⁾

Most participants reported that sexual reoffending was not a problem. More specifically, six months after treatment, only 6.2% reported that their sexual behaviour was a problem. This decreased in years one and two (5.2% and 2.5% respectively) but in year three, the percentage reporting that their sexual behaviour was a problem jumped to 14.3%. It then dropped dramatically in year four, with no participants reporting a problem.

Figure 1 presents the results related to relapse prevention. A maintenance plan and list of early warning signs were compiled for each sex offender before the program ended. Early warning signs include low self-esteem, feeling they deserve to offend, loitering and unexpressed anger.

Figure 1

Six months after treatment, 22.9% could not name the early warning signs leading to their offending behaviour. In years one and two, this dropped to a low of 10%. In year three, however, the percentage who could not name their early warning signs increased again to 17.9% then decreased slightly in year four

As shown in Figure 1, the percentage of sex offenders who said they used their maintenance plan decreased steadily during the first four years following treatment, from a high of 72.9% six months after treatment to 50% four years after treatment. At the same time, however, the percentage of sex offenders who reported being in problem situations (in which their maintenance plan would be required) decreased from 72.9% six months after treatment to 50% in year four.

Canadian Study

This research examined the long-term recidivism of child molesters who were treated for their sexual offending between 1965 and 1973. Comparisons were made between this treated group and two control groups of sex offenders who were sentenced to the same provincial institution but who did not receive specific treatment for pedophilia.

The comparisons between the treatment and control groups will not be discussed here, since this article is examining sexual recidivism in general. In addition, this study covered many other areas than can be reported here.

Method

All the men in this study had been sentenced to between 3 and 24 months for a sexual offence against a child. For the treatment group, information was collected directly from the offenders as well as from institutional files. For the control groups, information came directly from institutional records.

The treatment group and one control group had all been serving sentences for a sexual offence at the same time and at the same institution. The other control group served time at the institution before the other two groups.

About one third of the two control groups (32% and 35%) and two thirds of the treated group (63%) had previous sexual convictions.

Sexual offence recidivism, and not general recidivism, was the focus of the study. It was defined as a reconviction for a sexual or violent offence, as indicated by Royal Canadian Mounted Police records. Convictions for assault were included since it is common for sexual assault charges to be reduced to common assault through plea bargaining. Records for most of these offenders were obtained between 1989 and 1991. Because of missing information, the records from between 1974 and 1976 were used for 13 subjects.

Results⁽³⁾

Of the total sample of 197 child molesters, 42% were reconvicted for a sexual (or assault) offence during the follow-up period. However, the length of the follow-up period for the various groups differed depending on when the offenders were released. For example, on average, the follow-up period for the treatment group was 19 years, 28 years for one control group and 20 years for the other control group. This means that some groups were at risk for a reconviction longer than others. Using a statistical procedure called "survival analysis," the different lengths of follow-up (that is, at risk) periods were controlled, and this gave a new recidivism rate of 50.3%.

Figure 2 shows, for each year during the follow-up, the proportion of those at risk who were reconvicted for a new sexual offence. The rate of reconviction was 5.2% per year for the first six years. It then dropped to 1.8% per year for the next 20 years. Of particular note, however, almost one quarter of those who reoffended were reconvicted more than 10 years after being released from prison.

Figure 2

Looking at risk predictors that could be obtained from the offenders' files, the study found that offenders were at higher risk for recidivism if they had never been married, had prior sexual convictions,⁽⁴⁾ admitted to many previous sexual offences or had male victims. Offenders against males were at significantly higher risk for recidivism than the incest offenders and the offenders against females.

Variables that were unrelated to sexual recidivism included: age of the victim, a history of exhibitionism, a history of having been sexually victimized, a poor relationship with one's mother, alcohol or drug use, previous non-sexual convictions, age upon release, education and IQ.

Offenders who had a poor relationship with their father and who had extra familial (as opposed to intra familial) female victims were at a marginally higher risk for sexual recidivism, but these findings were not statistically significant.

The Risk Checklist

Among the many analyses done with these results, one looked at the combined predictive ability of the variables associated with a higher risk of recidivism. These variables were combined into a type of risk checklist. Each sex offender was given a score based on the following point scheme: unmarried - 1, married -0; only male victims - 2, intrafamilial female victim - 0, other victims - 1; two or more previous sexual convictions -2, one previous sexual conviction - 1, no previous sexual convictions -0. Each offender's score was then added up to determine a risk rating.

There was a strong association between the risk rating and eventual reconviction. With the exception of the two lowest categories (ratings of 0 or 1), as the risk ratings increased so did the recidivism rates. Moreover, if the risk rating scale was used to determine which of two randomly selected offenders (one eventual recidivist, one not) was going to be reconvicted, there was a five-in-seven chance of correctly identifying the recidivist.

Discussion

In the first study, we found that year three was the critical year for problems reoccurring. This was the year when the sex offenders reported the most problems. An analysis of the findings indicated that favourable changes occurred in the sex offenders' interpretation of their interpersonal, social and sexual adjustment and remained significantly changed, during the other years.

Furthermore, as behaviours improved, the sex offenders felt less need to use a maintenance (relapse prevention) plan, but their ability to notice the early warning signs that might lead to sexual reoffending continued to improve.

In the second study, we find support for previous research showing that child molesters are at risk for reoffending for many years. The greatest risk period appears to be the first 5 to 10 years following release, although almost one quarter of the recidivists were reconvicted more than 10 years after being released.

This study also confirmed several risk indicators that have long been identified as important in predicting recidivism among child molesters: previous sexual offences, never being married and the type of victim. The problem is that these risk predictors are fixed; that is, we are now looking at them after the fact, when it is too late to change them. Unfortunately, none of the changeable variables examined in the study (not all of which were discussed here) were associated with recidivism.

The results of these two studies present at least two challenges for future research. One is to identify risk indicators that can be addressed through treatment. The other is to ensure that, when examining recidivism and treatment outcome among sex offenders, we use long follow-up periods.

S. Margretta Dwyer and B.R. Simon Rosser, "Treatment Outcome Research:Cross-Referencing a Six-Month to Ten-Year Follow-Up Study on Sex Offenders," Annals of Sex Research, 5 (1992): 87-97.
R. Karl Hanson, Richard A. Steffy and Rene Gauthier, "Long-Term Follow-Up of Child Molesters: Risk Predictors and Treatment Outcome," User Report No.1992-02 (Ottawa: Solicitor General Secretariat, 1992).

---

⁽¹⁾After the interview, participants were also given a battery of tests to fill out and mail in. The results of these tests are not reported here.
⁽²⁾Due to space limitations, this article does not present the study's results on difficulties in the family of origin, work-related difficulties, difficulties with the offender's partner or wife and the offender's perception of his need for further treatment.
⁽³⁾Except where indicated, the analyses will focus on the 106 of the 125 treated offenders for whom recidivism information was available.
⁽⁴⁾However, the recidivism rates of offenders who had only one previous sexual conviction and those who had more than one were not significantly different.