Home‎ > ‎

Lösel and Schmucker again

posted 23 Feb 2017, 08:21 by Robert Forde   [ updated 23 Feb 2017, 08:54 ]
Elsewhere on this site (see "The effectiveness of offending behaviour programmes") I gave details of a 2005 review by Lösel and Schmucker, which claimed to find evidence of effectiveness for sex offender treatment programmes. There appeared to be numerous flaws in that review, which rendered the claim unsustainable.

In a more recent review, involving stricter criteria, the same authors reviewed sex offender treatment programmes again. This time they found no evidence for effectiveness in programmes that were conducted in prison. For programmes conducted in the community they found a small effect of about 6%. The authors suggested that some prison-based programmes did show some effect; it was the overall effect which was zero.

There remains a difficulty: if the overall effect was zero among prison-based programmes, but some did better, then logically some must have done worse. It is natural to find that there is variation in the data when conducting a study of this sort, but it is a mistake to conclude that the "better" studies were effective and the "worse" studies can somehow be explained away (with the implication that they weren't being run properly and might have done better if they were). For the same reason, the fact that some community-based programmes appeared to show a treatment effect may also be unreliable.

The authors noted that there were very few randomized controlled trials (RCTs) among the studies which they included in their review. As noted elsewhere, RCTs provide the strictest test of any treatment, and are mandatory in other fields such as medicine and pharmaceutical treatments. None of the RCTs in this review showed any treatment effect. Furthermore, as with their 2005 review, studies only showed a treatment effect if they were evaluated by the people conducting them (that is, if they were "marking their own homework"). Again as with the earlier review, studies only showed a treatment effect if they involved small numbers (less than 50). Again, this is classically what one would expect if the results were affected by a "bound for publication" bias.

Altogether, although the authors describe the results as "promising", it is very difficult to see how this study provides any support for the use of sex offender treatment programmes based on cognitive-behavioural approaches.These programmes have been undergoing evaluation, and showing allegedly "promising" results for 20 years or more. One may reasonably ask how long we have to wait before this promise is fulfilled, or we are entitled to decide that such programmes are ineffective. The truth is that when high quality studies are conducted there is no treatment effect.

Given the costs of these programmes to correctional systems throughout the Western world, and the fact that they tend to starve other approaches of funding, is it time for a change of policy?