The Study in Context  

[Page 371]

There is a widespread assumption in the mass media and probably amongst the public in general that sex offenders (especially those who offend against children) are particularly prone to repeat their crimes ( Grubin 1998: 32) . 

This assumption has also recently found its way into the Halliday review of the sentencing framework, Making Punishments Work, which states that: 'Sex offenders as a group may pose higher risks of reoffending' (our emphasis) (Home Office 200 1: 33). This may or may not be true, for there are no reliable studies of reoffending rates, but as far as reconviction is concerned the findings of this study reinforce those of other follow-up studies, all of which have demonstrated that the probability of a serious sexual offender being reconvicted for a sexual (and also for a serious violent) offence is relatively low, even for those who have victimized children and have been at liberty for a considerable number of years (Grubin 1998: 32-40; Thornton and Travers 1992).

It is recognized, of course, that the frequency with which sex offenders are reconvicted is not the only concern. Especially salient for sex offenders is the degree of harm inflicted. In other words the greater the harm, the lower tolerance there will be for accepting risks 

[Page 372]

(Monahan 1996), Obviously, the longer the period of follow-up after discharge from prison the more the proportion reconvicted will mount up, But the longer the period before reconviction occurs the less valid it will be to say that a 'mistake' was made if the risk were assessed as 'low' at the time when parole was considered, for so many unforeseen factors may have intervened to increase the risk of reconviction since the assessment was made, especially in the period beyond the possible parole licence. 

It appears that most follow-up studies of sexual offenders have not reported either on the seriousness of the sexual or violent offences for which they are reconvicted, or on the amount of time between the first reconviction and release from prison,

This relatively small-scale research project grew out of a study of parole decision-making in England and Wales carried out in 1992,1993 and 1994, Data were collected on more than 900 prisoners serving (determinate sentences of four years or longer who were being considered for parole in those years (Hood and Shute 1994, 1995). 

The sample included 192 male prisoners who had been convicted of a serious sex offence. Nearly half (46 per cent) were at their last possible review, about a year before they were due to be released at the two-thirds point of their sentence, The others were at earlier reviews: at the time the sample was drawn prisoners were eligible for parole consideration after having served one-third of their sentence, [*1] We were able to obtain from the Offenders Index all known reconvictions relating to 174 of these sex offenders, [*2] and from the Prisoners Index their date of release from custody . [*3]

[*1]  The policy, at the time this study was conducted, was that prisoners who were not recommended by their Local Review Committee and had a Reconviction Prediction Score (RPS) of 46 percent or above were not considered by the Board unless they were at their last review. As few sex offenders have such a high RPS this policy is unlikely to have skewed the sample of cases observed so as to exclude many high-risk' cases.

[*2] Eighteen offenders could not e traced. A comparison of some salient characteristics of those who could not be followed-up with those who could be traced showed that they were not atypical in relation to variables that the literature on sex offenders suggest are associated with reconviction.

[*3] All the prisoners were dealt with under the 'old' pre-1991 Criminal Justice Act parole system because they had been sentenced before the implimentation of the Act on 1 October 1992. For those who were not paroled at the review we observed we were able to assess whether they were paroled at a later hearing by comparing their actual date of release from prison with the two-thirds date (the 'earliest day of release' or 'EDR'). Where we had not been supplied with the date of release the Parole Board kindly checked the recommendation made by the Board at the prisoner's last review. There were only a few cases where we could not determine whether parole had been granted. 

By June 2000, most of these prisoners had been at liberty for a considerable length of time: 173 could be followed-up for two years, 162 for four years, and 94 for as long as six years, Thus it was possible to follow-up a substantial proportion of the cases for longer than the usual period of four years (see Friendship and Thornton 2001; Marshall 1994; Thornton and Travers 1992) which was a considerable advantage given the known initially low and relatively slow accumulation of sexual reconvictions over time. Indeed, it has been suggested that a follow-up period of at least five years for sex offenders is desirable (Lloyd, Mair and Rough 1994; Hanson 1998) .

Because these cases were collected as part of a decision-making study, the authors had access to the parole dossier for each prisoner. The dossier contained (in a police report, court transcript, social inquiry report or some other report) information about the nature of the offence(s) committed, This information made it possible to draw a number of distinctions between types of sexual offender, instead of treating them (as some studies have been forced to do) as an undifferentiated group. Thus, we distinguished between: 

[Page 373]

Between 1992 and 1994, 40 Parole Board panel meetings were observed by the authors (RH and SS), at 24 of which the sex offenders in the sample were being considered for parole. Twenty panels included a psychiatrist and these 'psychiatric panels' dealt with 90 per cent of the cases in this study. [*4]

[*4] In 1992 panels had four members, and in 1993 and 1994, three members. 

At every panel a verbatim record was made of the discussion. This made it possible, in most cases, to identify prisoners who were described by at least one member of the panel as posing a 'high risk', even if other members disagreed, which they occasionally did. Usually it was not difficult to identify such cases because the language used was unambiguous. Of course, the degree of risk varied, but cases were only classified as 'high risk' when at least one of the following, or similar, terms was employed: 'he is risky', 'a very high risk', 'a man of violence', 'he scares the living daylights out of me', 'alarming', 'frightening', 'an entrenched sex offender', 'an extremely grave risk to children', 'dangerous'.

On this basis, half of the cases fell into the 'high risk' category and, of these, 40 percent were at their last review and 60 per cent at a review earlier in their sentence. Of those not at their last review, at least 55 per cent were judged to be 'high risk' (in some cases no judgement had been made), compared with 44 per cent of those who were being considered at their last possible review, indicating that Parole Board members were probably more inclined to indicate concerns about 'high risk' when the period available for parole was longer. Where there was no overt mention of 'high risk' or 'danger' it did not mean that the prisoners were regarded as posing no risk at all. As one Board member put it (see Case F below): 'Obviously there is a risk, there always is.'

Thus, it was possible to show the extent to which members of the Parole Board were able to identify as 'high risks' those prisoners who subsequently were convicted of a sexual crime, or of a violent crime of such seriousness that it led to imprisonment (these we refer to subsequently as 'serious' offences). Of course, we only had knowledge of cases where a conviction was secured. We cannot, therefore, say how well Board members were able to predict sexual reoffending (on this, see below at page 393). Bearing this in mind, however, we were able to calculate the number who turned out to be: 

[Page 374]

The data pertaining to Parole Board members' identification of 'high-risk' sex offenders and the extent to which their assessments were later supported by the evidence of reconviction for a further sexual or serious violent offence are given added salience by the proposal in the Halliday report, Making Punishments Work. This recommended that there should be a special sentence for 'dangerous' sexual or violent offenders and that these 'dangerous' offenders should be the only prisoners subject to Discretionary Conditional Release (parole). 

If adopted, Halliday's proposal would mean that those deemed to pose a high risk of serious reoffending could be kept in custody until the full sentence had been served, rather than released automatically at the two-thirds point as is currently the case. 

In addition, the report proposes that the court should have the power to order sexual offenders to undergo a further ten years of supervision after the expiry of their sentence (Home Office 2001). 

Some of the prisoners in this study (particularly those with a previous conviction for a sexual or violent offence) would surely have been candidates for the new type of sentence, had it been available. We hope that our analysis of the extent to which the assessment of 'high risk' or 'danger' made by Board members correctly identified those sex offenders who turned out later to have been reconvicted of a sexual or serious violent offence will prove useful when the recommendations of the Halliday report come to be considered in detail.