From the Herald Sun of Australia comes an article that begins with this passage:

SEX offenders who commit further crimes are a small minority who can cause immense harm, according to a new study.

But the study concludes that the task of reliably identifying potential repeat offenders is "extremely difficult, if not impossible".

The research paper on recidivism of sex offenders found only 13.4 per cent were known to have committed a new sex offence within five years.

But Dr Karen Gelb, senior criminologist for the Sentencing Advisory Council, said the recidivism rate -- like the sexual assault reporting rate -- was likely to be a conservative estimate.

The opening sentence is a claim we often hear, but does it really fit with the actual data in this study? Considering the consequences, is "only" 13.4% accurately characterized as a "small minority," even on its face? That seems to me to be a quite sizable minority. More importantly, though, if 13.4% represents the fraction who reoffend and whose new offenses are reported and who are caught and who are convicted, then what is the fraction who simply reoffend? If X * Y * Z * W = 0.134 and if Y, Z, and W are all less than 1 (and Y, we know, is substantially less), then X may not be a minority at all.

Dr Gelb said Australian Bureau of Statistics figures showed that the proportion of sex offenders who moved all the way through the criminal justice system represented "only the tip of the iceberg".