![Sample our Geography journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5937/TOC-Subject-Sample-2021-Geography.jpg"})
Abstract
Citizens of developed societies are troubled by those who commit ‘irrational' crimes against the person. Reoffending by ex-patients following their release from secure mental health services triggers particularly intense angst when amplified by media and political scrutiny. Forensic mental health service providers are expected to minimise the occurrence of such transgressions by releasing only those patients who are judged acceptably unlikely to reoffend. However, reoffending probabilities can only be estimated by observing behaviour in secure institutional settings designed specifically to prevent patients from transgressing. The article explores this ‘inductive prevention paradox' which arises when the implementation of measures designed to avoid an adverse event obscures direct observation of what might have happened if prophylaxis had not been attempted. The analysis presented draws on data obtained in 1999–2003 from two qualitative studies in medium to low secure UK institutions, one providing forensic mental health services and the other forensic learning disability services. We explored the views of 56 staff members and 21 patients about risk management in forensic services and undertook additional 25 staff interviews for case studies of the 21 patients. The wider applicability of the inductive prevention paradox will be considered in the Discussion. We argue that the prognostic limitations arising from prevention have been underestimated by policy makers and in official inquiries; and that the prevailing personal risk assessment framework needs to be complemented by greater attention to the environments which patients will be discharged into.
Notes
1. The ‘inductive prevention paradox’ should be distinguished from the widely discussed ‘prevention paradox’ (Rose 1981) which arises from attributing a risk associated with an aggregate category such as alcohol consumption to individuals who meet the specified criteria for category membership. Both paradoxes are bound up with the limitations of probabilistic thinking. But the latter, which could be termed the ‘ecological prevention paradox’, derives from the requirement to shift between the aggregate and the individual in order to quantify probabilities (Heyman et al. 1998; Hunt 2003). The inductive prevention paradox results from limitations in the observational evidence base itself.
2. The extent and nature of media crime coverage varies across countries. A comparative study of 11 countries (Walgrave and Sadicaris 2009) concluded that the proportion of national TV news time devoted to this topic was positively associated with higher crime rates, greater media fragmentation and competition and stronger political polarisation. The UK was found to have some of the highest percentages of TV news items concerned with crime, nearly double those for France, together with a particularly strong slant towards personalistic accounts of violent offending. (The USA had an unexpectedly low rate of crime coverage which, the authors suggest, may have resulted from selecting national rather than more frequently used local TV stations.) A febrile media culture, reinforced by the predominance of crime as a TV drama theme and political competition to be ‘tough on crime’, leaves UK forensic mental health services facing a particularly difficult risk management dilemma.
3. Tools direct selective service attention to particular adverse events, in this case the risk of reoffending as against, for example becoming depressed or being attacked by members of the ‘community’. They thereby carry implicit value judgements (Heyman 2012).
4. Probabilities can only be quantified in relation to a temporal horizon, in this case three years, beyond which adverse events are not taken into account. Practically focused risk managers tend to frame time unreflectively.
5. More accurately, the probability of a patient reoffending is related to recorded history. Patients can influence their ‘history’ in this sense, for example by concealing previous offending. However, once their offending history has been encoded in a patient record, patients are cannot change it, unless they can demonstrate their innocence, a very unlikely possibility for those who have committed offences against the person.
6. Discharged patients may be directed away from localities associated with former offending. However, they will thereby also be separated from social networks and familiar surroundings. They tend to be discharged into areas of serious socio-economic deprivation which are associated with additional problems such as high crime rates and drug problems.
7. The concept of risk ownership originated in corporate governance where it is used to convey a top–down model of social order in which a senior manager at board level is made accountable for each risk which the organisation is deemed to face. As with the idea of risk management, the notion can be applied more generally to everyday life. For example, in the forensic mental health sphere, close relatives may decline to take back responsibility for a discharged offender (Heyman et al. 2010, p. 34).
8. This limitation applies to probabilities derived from correlations as well as to those in which a risk factor is put into a small number of categories, e.g. to both the relationship between number of cigarettes smoked and the chance of developing lung cancer and the relative risk for smokers versus non-smokers. In relation to the former, probabilistic inductive inference can only be applied to individual cases by ignoring deviations from the regression line. By extension, the same constraint applies to multiple regression unless perfect predictability can be achieved, and probability replaced by certainty, and projected randomness by determinism.
9. A parallel debate has taken place in the criminal justice system, with advocacy of a ‘tool’ for assessing reoffending risk, the Psychopathy Checklist–Revised (PCL-R) challenged by sceptics (Gendreau et al. 2002).
10. As this example illustrates, the ‘lens of risk’ may give a new frame for interpreting the social organisation of patient hospital stays (Roth 1963), in which the passage of time is seen to somehow reduce riskiness.
11. Randomised controlled trials (RCTs) can provide counterfactual evidence at an aggregate level. For instance, patients in the placebo group can be expected to have done as well on average as those in the treatment group if they had received an experimental drug. However, this aggregated knowledge offers limited predictive accuracy in individual cases, and the methodology of RCTs is difficult or impossible to apply with respect to long-term, complex interventions.