CR 168: Repeat Drink Driving Offenders in Western Australia, 1984 to 1994 (1996)


The aim of this study was to determine patterns of drink-driving arrests in Western Australia, and the characteristics of repeat drink-drivers. For the purpose of this study, drink-drivers were defined as those arrested for a drink-driving offence between 1984 and 1994. Repeat drink-drivers were those re-arrested for a drink-driving offence during the study period.


The Crime Research Centre at The University of Western Australia maintains a database of police apprehensions which contains demographic details of the alleged offender, identity checks, details of the charge (including date, place and nature of the alleged offence) and information describing the arrest process (that is, date of arrest, place of arrest, custody or bail arrangements). Demographic details were limited to gender, race (Aboriginal or other), date of birth, place of birth, and occupation.

Over the 11 year period from 1984 to 1994, 597,637 arrests were made by the Western Australian Police Service. Drink-driving arrests, that is, arrests that included at least one drink-driving offence (as specified above), comprised 133,599 or 22.4% of all arrests. A total of 127,225 individuals were arrested for drink-driving offences.

Methods of Analysis

First, a cross-sectional analysis of drink-driving offences was carried out, defining trends, rates, and demographic characteristics of persons arrested for drink-driving. There were rather more arrests than persons arrested, as each year there were a number of persons with multiple arrests.

The second stage of analysis consisted of a longitudinal study of the offending patterns of drink-driving offenders. A ‘criminal career’ model was used to explore various aspects of offending and re-offending - namely the interaction between age and offending (both age of onset of drink-driving offending and the peak and median ages of drink-driving offending were considered); frequency of offending; the patterns or ‘types’ of criminal careers; and the extent (if any) of specialisation or escalation in the types of offences committed by drink-driving offenders. An important aspect of any longitudinal criminal research is the temporal ordering of arrest events, from the time of first (ever) arrest, for each offender.

However, the data in use were ‘censored’, meaning that the complete criminal careers of each offender was not known because the study cut-off date was 31 December 1994. Without consideration for the effects of censoring, estimates of re-offending are likely to be seriously biased because the follow-up times for each offender vary. For example, a person arrested on the cut-off date would have had no opportunity to be re-arrested or establish any kind of criminal career. Data from a single year (1985) were used to provide some (crude) control of follow-up time, as this provided a follow-up period of nine years. The year 1984 was not suitable because of under-enumeration of cases for that year. This approach, while providing maximum follow-up time, markedly reduced the number of cases available for analysis.

The third and most sophisticated approach involved the use of a statistical method known as failure or survival rate analysis to determine the re-arrest probabilities of drink-driving offenders. Using a parametric statistical model known as the Weibull mixture model, the ultimate probability of re-arrest for all drink-driving offenders and for subgroups of the offenders arrested for drink-driving offences were estimated. In addition, a method of co-variate analysis was employed to test for significant differences in the probabilities of re-arrest between groups.


There are 10,000 to 12,000 arrests for drink-driving in Western Australia each year. These account for about one quarter of all arrests made by the police. For males, the rate of drink-driving arrests fell over the period examined but rose for Aboriginals, and was generally constant for females. About 45% of those arrested were under 25 years and about 60% were aged under 30 years. About 13% were female, about 66% had blue collar occupations, and about 10% were Aboriginal.

First time offenders, that is, individuals not previously arrested by police for any offence, made up two thirds of all drink-driving arrests. These first time offenders were much less likely to be re-arrested for a drink-driving offence or for any other offence, and also tended to be older at first arrest.

About 40% of arrests were for Driving Under the Influence (where the driver had a BAC of 0.15% or greater, or was deemed to be incapable of proper control of the vehicle), and 55% were for Excess 0.08% offences.

After nine years, about 70% of all offenders arrested in 1985 had only one drink-driving offence and about 20% had two offences. Around 80% of females had only one drink-driving arrest. Aboriginals tended to have a larger number of offences; one quarter had two offences, and 10% had four or more.

About half of drivers first arrested for any offence in 1985 combined drink-driving offences with arrests for other offences. About 30% of the 1985 drivers had multiple drink-driving offences, of whom two thirds also had arrests for other criminal offences.

Using failure analysis and the whole data set, rather than just one year of data, took into account the effect of different periods of follow-up available and confirmed the above findings. For all drink-driving offenders the probability of re-arrest for any offence was 0.47, and for a drink-driving offence, 0.32. These risks are lower for women (0.37 and 0.22). They are higher for males, Aboriginals, and for those with prior arrests. First time drink-drivers are much less ‘criminal’ than other offenders, while drink-driving offenders with prior arrests are considerably more so. For about 70% of those arrested, the drink-driving offence is their first and only arrest for any offence.

Repeat drink-drivers, compared with drivers with single drink-driving offences, were younger (65% less than 25 years v 45%), fewer were female (10% v 16%), more were Aboriginal (4% v 2%), and a greater proportion had the more serious, DUI offences (43% v 39%). Two thirds had other criminal offences. Aboriginals were over-represented in this group and tended to have greater numbers of arrests than non-Aboriginals.


The best predictors of repeat arrests for drink-driving were being male, under 20 years of age, having prior arrests, and being Aboriginal. These characteristics define a group for whom it may be difficult to define effective countermeasures. It may be better to concentrate on the larger number of offenders with single drink-driving offences, particularly those with the higher BAC offences. There is some evidence that there was a change in behaviour of these offenders after the introduction of random breath testing in 1988.

Unfortunately the arrest record does not have any information as to whether the arrest event was associated with a road traffic crash, therefore we are not able to say anything about the crash records of drivers in this study. The use of this set of data could be extended by linking the arrest records with traffic infringement data and with crash records. After obtaining the appropriate permissions it would be possible to define other important driving and crash-related characteristics of drink-driving offenders. These data could also be used to evaluate the effects of different countermeasure programs.

Download Complete Document: Alc_Repeat [PDFPDF: 860 KB]

Type: Research and Analysis Report
Sub Type: Consultant Report
Author(s): Ryan, Ferrante, Loh & Cercarelli
Topics: Alcohol
Publication Date: 01/10/96