Retail Drinks Australia have published to members an in-depth research project that first started in late 2022 by Data Analysis Australia called Liquor Outlet Density, Accessibility & Harm

Modelling the relationship between the availability of packaged liquor to community harm is a simple problem that manifests statistically difficult.  The difficulties involved include the observational nature of the data, the need to consider other factors related to harm, the inherit spatial nature of the data with irregular geography, spatial dependence and time series aspects.  The implemented methodology uses a Point-to-Area approach to handle distances between outlets and Statistical Area 1s, using road distances from OpenStreetMap.  This approach created the best approximation of distance of a liquor outlet to an analysable area. Liquor outlets effects were fit with quadratic splines assuming a positive relationship between outlets and harm like many in the field have theorised. This resulted in a statistical framework that combined algorithms for managing generalised linear models that reflects the count nature of the data on harm, and the spatial and temporal dependence structure required for a longitudinal geographical analysis.  

High level findings indicated socio-demographic and socio-economic effects explained a large proportion of the geographical variation in harm, and different types of liquor outlets behaved differently.

The findings and methodology show how comprehensively and robustly statistically modelling a scenario like this leads to understandable, explainable and analysable results – fit for any funding, research or legal matter.  This project is another example of how Data Analysis Australia design bespoke models and solutions for each of our clients.

The work was led by Dr John Henstridge, and DAA thanks the co-authors of the work Elyse Corless, Dr Fiona Evans, Anna Hayes and Lachlan Robinson for their great work.