The Problem 

Western Australia has, in common with a number of other states, a system to confiscate the proceeds of criminal activities. This money is then used for crime prevention. 

The Attorney General made accusations that the Confiscation Proceeds Account Committee (CPAC) and particularly its Chairman, member for Perth, were responsible for maladministration and political manipulation of funding grants from the Confiscation Proceeds Account. These claims were based on an apparently disproportionate allocation of funds to Labor electorates and in particular to the electorate of Perth. As a result of these claims, a Parliamentary inquiry by the Public Accounts Committee was launched into the distribution of grant funds. 

Crime data is always used as a basis for the CPAC to make determinations, so part of the enquiry assessed whether the disproportionate allocation of funds to Labor electorates was comparable with their crime rates. As crime data is not readily available at the electorate level, the challenge was to compile data from smaller geographical regions, some of which cross multiple electorate boundaries, to produce estimates of crime in each of the Western Australian electorates. Data Analysis Australia was asked by the researchers of the Parliamentary Library to produce these estimates for 2003, 2004 and 2007, chosen as they were years in which funds were distributed and no electoral boundary changes occurred. 

The Data Analysis Australia Approach

 Western Australian Police (WAPOL) statistics are readily available for each of the offence localities in Western Australia for six offence groups: assault, burglary (dwelling), burglary (other), motor vehicle theft, robbery and graffiti. By aggregating offences at the locality level it was possible to estimate offence numbers at the electorate level. 

Using the appropriate State Electoral Boundaries, the first step was to allocate offence localities to state electoral divisions (SEDs). Some localities fell wholly within an electorate but others did not. When a locality overlapped two or more electorates, the offence data for that locality had to be apportioned between electorates. 

From the Australian Bureau of Statistics' Census of Population and Housing it is known how the population of overlapping suburbs is distributed between electorates and, where appropriate, this proportion was used to distribute offence counts. As offence localities include amenities, regional landmarks and so forth, it was necessary to allocate some localities manually. This was done using maps of the state electoral boundaries. In the rare case where a locality fell in more than one electorate, proportions were estimated based on the proportion of the locality in each SED and the size of Census Collection Districts in the area. 

 The data was then aggregated for localities and part localities within electorates and offence counts and rates for each electorate were calculated. Offence rates were estimated using population counts for that electorate based on the Estimated Resident Population, which is based on Census data and produced by the Australian Bureau of Statistics. This step of taking electorate populations into account is important as it provides a more comparable estimate of crime issues than considering the offences themselves, as one could expect more offences in more highly populated electorates. 

The Result

 The resulting estimates of offence counts and rates at the electorate level enabled a summary of electorates by party to be determined. Key outcomes included:

• For the three years examined, the electorate of Perth had the highest crime rate by a considerable margin;

•  In 2003 and 2004, Labor electorates comprised eight of the 10 electorates with the highest crime rate in the state; 

• In 2007, Labor electorates comprised nine of the 10 electorates with the highest crime rate in the state; 

• In 2003 and 2004, only one Labor electorate out of a total of 57 electorates fell within the 10 electorates with the lowest crime rate; and 

• In 2007, two Labor electorates out of a total of 57 electorates fell within the 10 electorates with the lowest crime rate. 

It was concluded by the Public Accounts Committee that the allocation of funds to Labor electorates from the Confiscation Proceeds Account was comparable with crime data for those electorates. This report was published and tabled in Parliament. 

This approach is one that can be applied in many different contexts. In addition to ongoing crime statistics reporting, Data Analysis Australia also produces biannual unemployment statistics by electorate using the Department of Education, Employment and Workplace Relations' estimates of unemployment and the labour force for each Statistical Local Area across Australia as a basis. Data Analysis Australia has developed tailored automated reporting tools using the statistical package R to compute the estimates and produce the reports. Such automated reporting ensures consistent, timely and high quality output.  Click here to view unemployment statistics for your electorate.