A census of prison inmates has been carried out biennially in November since 1987. The Department of Justice administered the census until 1993, and the Ministry of Justice until 1997. The Department of Corrections took over this responsibility in 1999. The current census provides statistical data on inmates within the prison system as at 20 November 2003.

The census contains data on offender age, ethnicity, sentence details, offending history, gang membership, education, income, children and marriage, iwi affiliation, programmes and activities, use of medication, and other topics.

The purpose of the census exercise is to provide updated information concerning key characteristics of the New Zealand inmate population. Taken together with previous censuses, trends and patterns in this population may be identified.

Information was collected from several sources. For the first time, the census used the Department of Corrections’ Integrated Offender Management System (IOMS) database as the primary data source. To obtain details not normally recorded in IOMS, a supplementary census questionnaire (comprising a small number of questions) was then sent to prison staff to complete in interview with inmates. Prison health services staff also collated numbers of inmates on medication, and programme personnel collated numbers of inmates on programmes.

Inmate numbers tend to fluctuate in a predictable manner over the course of each week. The general pattern is mainly a function of prison releases occurring on Mondays, Tuesdays and Wednesdays. The prison population tends to decrease slightly over those days, and then gradually builds up again to a peak on Sunday nights. To be consistent with prior censuses the current census was scheduled for a Thursday. Home detention figures relate to the Sunday following the day of the main census.

The set of offenders summarised for the Home Detention section was extracted from the Department’s data warehouse (known as Corrections Analysis & Reporting System, or CARS), using a “snapshot” dated 23 November 2003, a date three days after the prison census data collection.

Where possible, time series data are presented to enable comparisons between current figures and those of previous prison censuses (often as far back as 1991). Unfortunately some of the time series comparisons are subject to discontinuities due to a change in census methodology from one year to the next. Changes specific to the current census’s methodology are discussed in detail in chapter 17 of this booklet.

As with any statistical data collection, the presence of error is to some extent inevitable, for a variety of reasons. While several strategies have been employed in this exercise to minimise errors and provide quality assurance to the data set, given budget and time constraints it must be accepted that some errors will have survived.

The methodology used in this census will be reviewed prior to the next census in November 2005, and further changes may be made. Further, there is likely to be further revision of the topic areas included (and excluded) from future censuses. Feedback regarding the current statistical data set, and comment about the changes made to content and format, are welcome.