Hmm....I had a look for the DoJ report I read earlier to get the real numbers, and that one seems to have gone (or I just dont remember it very well).
Anyway, I did find two others based on the same data, one dealing with recidivism among general prisoners released (
here), and another one dealing specifically with sex offenders (
here).
So, overall recidivism rates among violent criminals (homicide, rape, assault and robbery, but not including sexual assault for some wierd reason, which makes it hard to figure out the overall figures) after 3 years is 61.7%. About half to 2/3rds of that happens in the first year, depending on how many priors they had (Yikes!). Of those convicted of rape and sexual assault, 46% and 41.5% respectively will be rearrested for something (violent and non-violent) within 3 years.
Rearrest for rape (sexual assault rearrests weren't given for the general offenders report) over 3 years were 2.5% for rapists and roughly 1.1% for other violent offenders.
Overall rearrests among those originally convicted of sex crimes for new sex crimes is 5.3% for rapists and 5.5% for sexual assaulters. So, rearrests are roughly half for rape and half for sexual assault by extrapolation for the data in the general prsioners report. Given that, the probability of sexual re-offence by the general violent criminal should be around 2.2-2.5% (thereabouts). Conviction rates were approximately two thirds the arrest rate.
Of those re-offences, about half were against children. For rapists, the number was 1.4% and sexual assaulters 2.5% over 3 years.
Among those originally convicted of sex crimes against children, similar numbers for overall sexual re-offences apply (about 5%), but, re-offence against children was 3.3% over the three years. The balance of sexual re-offences took place against adults. So, the probability of a child molester re-offending over a period of 3 years is roughly about 1.1% per year. Probably that will drop off over time beyond the 3 years since those likely to re-offend would probably do it sooner rather than later.
The probabilty of re-offence against children is about three times higher for those with multiple convictions compared to those with a first offence (which appears to be the majority of the SOs released). In other words there is a fairly small sub-population that we would need to be concerned about, as they are the ones with a high probabilty of re-offence. Basically those who have been arrested multiple times.
Anyway, if you are interested in these things, read those two reports.