OBJECTIVE: The purpose is to answer the following research question: are the time-series data published by the Australian Bureau of Statistics for Queensland statistically the same as those of the Queensland Suicide Register? METHOD: This question was answered by first modelling statistically, for males and females, the time series suicide data from these two sources for the period of data availability, 1994 to 2007 (14 observations). Fitted values were then derived from the 'best fit' equations, after rigorous diagnostic testing. The outliers in these data sets were addressed with pulse dummy variables. Finally, by applying the Wald test to determine whether or not the fitted values are the same, we determined whether, for males and females, these two data sets are the same or different. RESULTS: The study showed that the Queensland suicide rate, based on Queensland Suicide Register data, was greater than that based on Australian Bureau of Statistics data. Further statistical testing showed that the differences between the two data sets are statistically significant for 24 of the 28 pair-wise comparisons. CONCLUSIONS: The quality of Australia's official suicide data is affected by various practices in data collection. This study provides a unique test of the accuracy of published suicide data by the Australian Bureau of Statistics. The Queensland Suicide Register's definition of suicide applies a more suicidological, or medical/health, conception of suicide, and applies different practices of coding suicide cases, timing of data collection processes, etc. The study shows that 'difference' between the two data sets predominates, and is statistically significant; thus the extent of the under-reporting of suicide is not trivial. Given that official suicide data are used for many purposes, including policy evaluation of suicide prevention programmes, it is suggested that the system used in Queensland should be adopted by the rest of Australia too.
OBJECTIVE: The purpose is to answer the following research question: are the time-series data published by the Australian Bureau of Statistics for Queensland statistically the same as those of the Queensland Suicide Register? METHOD: This question was answered by first modelling statistically, for males and females, the time series suicide data from these two sources for the period of data availability, 1994 to 2007 (14 observations). Fitted values were then derived from the 'best fit' equations, after rigorous diagnostic testing. The outliers in these data sets were addressed with pulse dummy variables. Finally, by applying the Wald test to determine whether or not the fitted values are the same, we determined whether, for males and females, these two data sets are the same or different. RESULTS: The study showed that the Queensland suicide rate, based on Queensland Suicide Register data, was greater than that based on Australian Bureau of Statistics data. Further statistical testing showed that the differences between the two data sets are statistically significant for 24 of the 28 pair-wise comparisons. CONCLUSIONS: The quality of Australia's official suicide data is affected by various practices in data collection. This study provides a unique test of the accuracy of published suicide data by the Australian Bureau of Statistics. The Queensland Suicide Register's definition of suicide applies a more suicidological, or medical/health, conception of suicide, and applies different practices of coding suicide cases, timing of data collection processes, etc. The study shows that 'difference' between the two data sets predominates, and is statistically significant; thus the extent of the under-reporting of suicide is not trivial. Given that official suicide data are used for many purposes, including policy evaluation of suicide prevention programmes, it is suggested that the system used in Queensland should be adopted by the rest of Australia too.