Reference projections represent the simulated description of what we could expect to see in the future based on current  trends .  The simulation engine incorporates major changes in the past 20 years to the working of the criminal justice system and the reference projections include a historical period. This, depending on the specific statistic being monitored, may be up to 20 years. The forward looking period is similar, up to the year 2040.  Where possible the historical period is also shown against published statistics recording particular aspects of the operation of the criminal justice system, for example the number of offenders in custody or under supervision in the community. 

These published 'reference data' provide one aspect of simulation validation, helping to assure that  the underlying simulation engine is capturing the essential dynamics sufficiently well to give a good representation of the past (at least in so far the published data are, in themselves, an accurate description of the state of affairs at the relevant time) .  The simulation engine represents, of course, a simplification of the real world system, and it is very important that it is capable of predicting the correct direction of change, even where the comparison with published statistics is not favourable. This gives confidence that assessing the impact of a potential policy or practice change will tell us whether it will represent an improvement or not, even if there is more doubt about the size of the effect. For the most part, however, simulations are proving robust with differences between simulated results and published statistics of less that 10%, often less than 5%.

The dynamics of justice systems are complex and non-linear, that is with feedbacks impacting on many timescales, months, years and decades. In consequence, some offender groups (e.g. those serving long or life sentences)  show very considerable "inertia" - it takes a long time for changes to work through - while others (e.g. those on short sentences or brief periods of community supervision) are more sensitive with impacts felt relatively quickly.   The results given above - and others like them - provide benchmarks for policy analysis and  illustrate some of the characteristics described above.  See the Policy Examples for a range of policy or practice assessments.


The projections generated by the simulation engine compare well with published statistics over the past 15 years.

Assuming current  trends for population changes, offending and sentencing practice continue, it is estimated that:

*  the male prison population will grow from around 83,000 in 2017 to about 90,000 by 2035, and
*  the post-release supervision caseload will grow from approx. 67,000 to 76,000 over the same period.

Similarly, community sentences and suspended sentences grow from 65,000 to 70,000 and from 38,000 to 43,000 respectively.

Reference projections

Top Point:

Assuming current trends in population growth, offending and sentencing, the number of male offenders in custody would grow from around 83,000 in 2017 to around 90,000 by 2035.  The numbers of those supervised by probation on post-release supervision grow from around 67,000 to 76,000 over the same period.
The application of scientific, quantitative methods to questions of policy and practice in criminal justice