As in the previous section, information on program outcome is based on financial year data provided by jurisdictions for the Report on Government Services and other unpublished data collected by NCAG. Additional breakdowns of data into front-end and back-end program numbers were provided directly for this study by jurisdictions that operate both types of home detention.
Jurisdictional comparisons of program outcome are particularly vulnerable to the small number effects discussed in the previous section. For example, ACT’s annual completion rates as reported in the RoGS have ranged from 50% to 100% since the program commenced. A single order breach reduced the completion rate from 100 in 2001-02 to 67% in 2002-03, and the 50% rate in 2003-04 reflects only two order revocations during the year. The same number of order revocations (2) produced a completion rate of 83% in 2004-05 because a larger overall number of orders was completed in that year than in 2003-04 (12 compared to 4).
In Victoria, the home detention program operates on only a pilot basis and commenced in January 2004. At June 2005, only 64 orders had been completed in total. ACT’s program has been operational since September 2001 (and was expanded to include home detention for unsentenced offenders in September 2003) but only 23 orders had been completed in total for both types of home detention up to June 2005. Because of the potentially distorting effects of these small number variations, the jurisdictional comparisons below exclude Victoria and the ACT from the analyses. Information from these two jurisdictions is included in the following sections, but presented for descriptive rather than analytical comparison purposes.
In SA, only 32 front-end home detention orders had been completed over the fiveyear period they have been operating. Given these small numbers, separate comparisons of front-end and back-end home detention across jurisdictions also exclude SA, although the information is provided for descriptive purposes only.
Other jurisdictional comparisons also need to be interpreted with caution. For example, in the NT, although information is available to this study on over 900 orders since 1993-94, the maximum number of orders completed in any single year is only 110. This means that trends based on annual statistics can also be subject to small number effects and year-by-year fluctuation, thereby limiting the validity of statistical comparisons based on yearly figures.
In effect, analysis of performance based on quantitative information will only provide meaningful interpretation where there are larger data sets. To compare performance variation between jurisdictions on outcome measures also requires a sufficient number of orders over time to determine a valid and stable level of performance. This is the case for NSW, Queensland, SA, and NZ where information on orders and their outcomes is available for thousands of orders. However, even in these cases, changes to policy, procedure and operational factors over the time period may affect completion rates, and therefore reduce the validity of jurisdictional comparisons based on a total order completion rate over a long period of time.