The first and most obvious thing to do is to look at those who do not reoffend and work out why, the reality is that these are more the exception than the rule.
Just to give you and idea how figures fro reoffending are manipulated,
In the UK we generally use some figure of reoffending within a set period of time, well we can choose that time period for our convenience, different agencies will use different time periods - so someone running a specific program will use one period of time, lets say 18 months, because it reflects well upon their program.
There are certain weaknesses, fro example different ages of offenders and different types of crimes have different reoffending rates, young people - below 21 - in the UK have a completely diabolical reoffending rate - you are looking at very much worse than 90% in a year, probably less than a year, this number is subsumed into the wider picture of all offenders.
Another weakness is that those who are newly released on licence are more closely supervised, especially those who are on electronic tags or gps tags - these individuals are less likely to reoffend in person (although they can still be part of offending by directing others) This tends to make tagging look good, but I have yet to see any figures for reoffending for those who have been taken off tagging. I can only speculate that there is a reason for this, that in the longer run tagging does not work.
There are a number of those who are very unlikely to reoffend, and it is obvious to everyone - the reason is that their offence does not lend itself to repeat - fine defaulters, breach of court orders, more serious traffic offences. It makes no sense to include these in the reoffending figures because they were not going to reoffend anyway and the prison system has had no impact upon their behaviour.
When you look at the longer term reoffending rates, the figures go up dramatically, I have seen figures for 18 months reoffending, and for 3 year reoffending which, but I have only rarely seen the figures for 9 year reoffending - this is not surprising because the 9 year figures are utterly appalling, I recall these numbers were well over 70%. If you look around though, you find it very hard to find reoffending figures for periods of more than a year - my view is because the reality is politically unacceptable.
There are organisations who do gather such figures, but generally they are not given any weight at all in developing policy, although I have seen them submit evidence to the many inquiries and white papers that the government uses as part of consultation.
Here is a lot of information in the link below, it does require some interpretation instead of taking it at face value - you need to be working within prisons to truly appreciate what it says - however I will draw your attention to page 5 of the .pdf which demonstrates all too readily how political interference has hamstrung the entire judicial system in the UK (and I won’t go into the ‘reform’ process in the courts system which will mean hundreds of courts being closed)
http://www.prisonreformtrust.org.uk/Portals/0/Documents/Bromley%20Briefings/summer%202016%20briefing.pdf
You should also look at pages 11 and 12, these relate to work activity in prisons - now I used to teach prisoners certain work skills - to the tune of 1500 level 2 qualifications a year, essential stuff for them because those qualifications related directly tot eh sort of work that prisoners are most likely to access - that training I did was withdrawn, I actually do another job that pays better.
The work activity in prisons is now largely meaningless, because it is only predicated upon the numbers of prisoners you can put into a workshop with as few staff as possible to operate the work areas - unsurprisingly the number of employment based qualifications has fallen off a cliff.
This is absolutely all down to money and political theory, nothing else whatsoever is responsible.
