This week I launched my newest open data project inspired by cashless income management and its relationship to crime rate data. The project uses QLD Police Service crime rate and socio economic data. The SEIFA data is more to provide insight into the areas under question. The crime rate data is useful in the arguments about whether or not certain communities should be placed on blanket income management because it is reported as a rate per 100,000 population which is meant to facilitate comparisons across the country.
Past forms of income management have applied specifically to individuals who have been found to have a need for income management and been referred for management by child services, for example. This contrasts with the new form of income management which the government seeks to apply to all people of working age in receipt of welfare, regardless of individual history. In contrast to dealing with people as individuals, the government seeks to justify blanket income management with reference to welfare payment and crime rates (although propaganda, innuendo and rumour often suffice).
It is in this context that I decided to examine the crime rates in QLD which contains the Hinkler electorate, this being the site chosen most recently for the imposition of blanket income quarantining. I would have liked to compare with Western and South Australian crime rates (the jurisdictions for the original Trial sites) but this data is not available in the same way. This is often a problem when trying to use public data for any meaningful purpose- there is no consistency in terms of availability, format, breakdown or timeliness.
For example QPS publishes crime rate by month on a calendar year by Region, District & Division, WA uses a financial year and totals by state and with broad groupings not each individual offence type. State laws vary so this complicates comparison of crime rate data across state boundaries. To get some idea of the number of data points that has to be managed in comparing crime data take a look at this helpful list of offences under Vic law. QPS data contained data points for over 80 offence types with some of those being sub-totals of other offences, making the job more complex.
I contacted the ABS for data which would serve the purpose of allowing me to compare the crime rate of one area in one jurisdiction with another area in another jurisdiction but was told that the ABS did not have such data and that I should try to contact various government departments to ask if they might have it- not a very helpful response and also brings into question what kind of data the government is relying on in making its decisions? Given my short time frame for the initial phase of the project, I settled on evaluating only QLD crime rate data which is published by QPS.
You may wish to try searching the Australian Government Directory at <directory.gov.au>. This website contains contact details for various government organisations, which may assist you in locating an alternative organisation with information relating to your enquiry. (ABS response)
It is also an issue when administrative boundaries for important data sets such as crime do not fit with any of the boundaries used by the ABS. This makes it very hard to match with other data or be 100% confident when drawing conclusions from one dataset to another (or simply usable in terms of searching/retrieving data in a project such as this).
Fortunately POLSIS created community profiles for each QPS District which distributed SEIFA and population data according to the QPS boundaries, allowing me to include this data in the project- although I had to extract this data manually and create my own CSV’s. While DSS does provide income support payment data by electorate and LGA, they obviously do not provide it for QPS boundaries so while I have used this data in previous projects I was not able to use it in this project due to lack of overlap between these types of areas and QPS boundaries.
One of the issues with open data projects is that they can become very confusing for users if there is too much drill down so I ended up locking down the option to search or drill down (ignoring about 50% of the code I wrote) until I could find the time to give the design the attention required to ensure that any drill down provided enhanced rather than compromised the user experience. The result of taking out all the search and drill down may make the project seem over-simplified in the short term but it was important for me to ensure the lessons from it are clear in my submission to the Inquiry into the Social Services Legislation Amendment (Cashless Debit Card) 2017 Bill which was due (and which I submitted) this past week.
The kind of information people often find the most meaningful, like being able to compare data (whether it be crime rates, alcohol sales or government funding) by area is often the kind of functionality which is most notably absent from public data offerings. The ABS will often only publish data at a state level without the kind of breakdown that would allow people to compare one area to the next.
When I present open data projects it is usually the first time that kind of break down and comparison has been made publicly available and this is potentially the first time this ranking has been created. I haven’t had the time to check to see what I can find in closed academic environments but it is obviously the case that if this kind of data were already publicly available, I’d have found it and used it rather than having to create it myself from scratch.
It would obviously be interesting to expand the project to rank administrative areas across every state/territory and not simply the state with the most raw data available in machine readable format (which fortuitously happened to be the state I was interested in drawing conclusions about). Long story short- it can be seen from the data that the Wide Bay Burnett District which relates to the Hinkler electorate does not support the notion of it being a QLD crime hot spot. Far from it, the Wide Bay Burnett District is in the lowest third of the 15 Districts.