Good statistical practice depends on the application of sound statistical methodologies. OESR provides assistance to agencies in developing such methodologies. We can design procedures and build tools tailored to the particular needs of your agency and your project.
To ensure improved use of administrative data, OESR develops a wide range of data cleaning and manipulation methods. For example, OESR has developed a name-matching package that allows data from numerous sources to be matched.
Many of the datasets kept by government use individual names as their primary identifiers; thus when data are poorly recorded (e.g. mistyped or misspelt names), it is difficult to track individuals through the dataset. The matching techniques developed by OESR allow for these records to be matched flexibly and automatically. This has been undertaken on data from the Queensland Criminal Courts Database, the Department of Communities and the Queensland Police Service, allowing previously impossible research into reoffending in Queensland to be conducted.
Classifications and standards
Delivery of high-quality statistics requires the use of agreed statistical standards. OESR has worked with Queensland Government agencies to develop standards, including classifications and counting rules. We also provide guidance and assistance on national standards, which includes developing mappings to standard classifications from existing operational classifications.
Extracting desired information from large databases may be a complex task. The difficulty may increase if visual inspection is sought to determine patterns within the data, or if statistical analyses of the data are required. OESR has developed user-friendly, menu-based data extraction tools enabling easy manipulation of data, including the ability to export data to software packages for statistical analysis. These tools permit data to be manipulated quickly and easily, with the added advantage of being distinctly separate from the database. The tools and data may therefore be used by people without requiring access to the database itself.
Last reviewed 20 December 2006