As I pointed out, in the big review by Marc Renzema, he whittled it down to three high-quality studies. What we find is that, yes, there could be evaluations out there, but most of the evaluations are problematic. Is it because the breaches are wrongly or poorly measured? Is it because one study uses one kind of offender—an impaired driving offender, let's say—while another study uses sex offenders? What's it due to?
I've seen no analyses that attempt to figure out whether the differences in breach rates are due to the type of offender or the policies in that jurisdiction for breaching someone.