The AIHW indicates that -
High quality health information systems are essential for the provision of high quality population health services. The Computer Assisted Telephone Interview (CATI) health survey is an example of a practical and efficient method to collect health information. To conduct effective CATI surveys it is necessary to have access to representative population samples; typically, this means access to all potential active household telephone numbers. The aim of this electoral roll matching project is to test an alternative sampling frame—the electoral roll—as a starting point for accessing telephone numbers.The report contains analysis of the match outcome and survey response outcome. What sample are we talking about in considering those outcomes? The AEC extracted 64,855 records) from the electoral roll at the request of ARCPOH. 496 duplicate records were removed from the AEC dataset based on matching surname, street address, suburb, postcode and state. This resulted in 64,359 AEC records being sent to Sensis for matching against the MacroMatch database. As a result some 1000 records containing data errors and 30,000 non-matched records were removed from the original sample.
The Australian Research Centre for Population Oral Health (ARCPOH), a collaborating unit of the Australian Institute of Health and Welfare (AIHW), requested the Australian Electoral Commission (AEC) to extract a sample from the electoral roll. These data were matched to the Sensis MacroMatch database to append a residential telephone number. The most complete matches were used as the sampling frame for the 2008 National Dental Telephone Interview Survey (NDTIS) conducted by ARCPOH. AIHW analysed these data using the following dimensions:• sex
• do not call register (DNCR) status
• main phone line type (fixed line or mobile) listed for the record
• region of residence (metro or non-metro).
The authors claim that -
* On average, just over half of the records (51.8%) were adequately matched between the electoral roll records and the Sensis MacroMatch database.They go on to suggest that -
* Matching rates were not consistent across all states and territories; for example, the Northern Territory had the lowest match rate for males (33.7%) and Tasmania had the highest match rate for males (59.5%).
* There were no substantial differences in match proportions between males and females nationally or among states and territories.
* Higher proportions of females completed the survey than males.
* Survey completions were influenced by the do not call register (DNCR) status: people who registered were more likely to complete a survey than people who were not registered.
* People living in the non-metropolitan areas completed the survey at a greater rate than their metropolitan counterparts.
The results in this report suggest that it is feasible to access telephone listings for surveys by using the Australian electoral roll databases as a starting point. This report enables states and territories to determine their status on match and response outcomes, which may be useful for planning future survey programs.A salient concern is the demonstration of the effectiveness of matching, a matching that might be undertaken by private sector organisations or that might be abused by government. That matching is, of course, already taking place.
Another salient concern is the absence of a reference in the report to privacy concerns: those concerns appear to be a non-issue.
