The importance of numbers: Using capture-recapture to make the homeless count in Adelaide

An important problem for the homeless service sector is understanding the size of homeless populations, which has implications on planning services and social policy. The aim of this study is to apply capture-recapture methods to count the primary homeless population in the Adelaide city council area, to examine the use of an alternative method to the Australian Bureau of Statistics census. Capture-recapture techniques were used to analyse homeless registers from three different services to estimate the number of primary homeless people in the Adelaide city council area from 19 June to 19 September 2005. Log-linear model and the sample coverage method were employed to analyse the data. The log-linear model results gave a population estimate of 455 (95% confidence interval 299, 762), and the sample coverage method of 311 (95% confidence interval 229, 466), compared with 104 from the Australian Bureau of Statistics census. Multiple sources of information utilising different methodologies should be considered together when attempting to plan services for primary homeless people, as all available techniques have important limitations. Capture-recapture is an important method to supplement any attempt at enumeration of hidden, mobile or difficult-to-reach populations.