Far too many folks try to use this Canadian Study (Sex Offender Recidivism: A Simple Question 2003-2004) showing 24% recidivism to refute the Dep't of Justice study which shows 3.5% recidivism rates.
Now the truth as to why it cannot be compared.
First the Canadian study is a Meta Study obtaining its input from other studies, and thats where the problem starts.
A review of Table-1 (pgs 2-4) (A list of the studies used as input) will reveal, 5 of the 10 studies used, include "Charges and Convictions" in their recidivism numbers, that alone will raise the output percentage. Then one of the studies uses "Investigations" of sex crimes, in addition to "Charges and Convictions" again raising their final recidivism percentages; its mathematically impossible to compensate for that.
Secondly, the Canadian meta study is using TWO studies from the 1980s, before any known therapy existed, obviously a high recidivism rate would be expected from that them. I'm sure if you combined "Charges and Convictions" in the DOJ study the final recidivism rate would be higher, but not everyone charged is convicted; in-between the two is where justice and truth is applied.
Accordingly, given the padding (as it were) of the input in the Canadian study, it can only arrive at a higher recidivism rate. This is like comparing rocks to eggs, they simply do not computer!
PS: Also, they did not follow folks for 15 years, each study followed folks for a different number of years, from 2 to 23