> Resources and Publications > PHC RIS infonet > October 2008

Editorial: How do you make data count?

Whether you've had a lot or a little to do with data, chances are you've heard the popular saying “lies, damned lies and statistics”, where even accurate data can be used to reinforce inaccurate arguments. In environments where statistics can influence funding and resource allocation, such as in health care, it is particularly important to make data count in the right way through accurate collection, independent analysis and unambiguous presentation.

The Annual Survey of Divisions (ASD) has evolved over the last few years to deliver the most representative Division data possible. A large part of this process involved the conversion of the survey in 2006 from a word document to an on-line, web-based system (the Divisions Information Online System). This system has greatly improved the accuracy and quality of the information collected with data checks built in to prevent missing data and non-meaningful responses (eg. the total full-time equivalence (FTE) of GP staff working in a Division cannot be greater than the number of GP staff). In addition, to minimise different interpretations, definitions of terms have been provided where appropriate.

Beyond automated data checks in the survey itself, a range of other methods are used to further enhance the quality of the data. If outliers (extreme values) are present or if substantial changes are observed in a value compared to the preceding year, these are double-checked with the Division. In addition extensive feedback about responses is provided to each Division allowing them to check and correct inaccurate information.

Data analysis and presentation of the results is completed objectively and thoroughly by PHC RIS. Analysis is repeated by a second researcher to ensure validity and results are supplemented with contextual information on relevant health and political issues.

There will always be some degree of variability in the accuracy of data reported but as long as we remove as much error as possible then we are making the most of the data. Information technology plays a pivotal role in achieving this. Non-statistical benefits attached to such a rigorous data process include reduced burden for Divisions and PHC RIS alike, with data anomalies prevented or corrected in a timely manner.

For a more detailed overview of the steps involved in gathering and reporting on data from the ASD, please go to <www.phcris.org.au/products/asd/results/06_07.php>.