Estimating audiences: sampling in television and radio audience research

Cultural consumption is problematic from a number of different perspectives, but certain responses from producers, regulators and commentators depend on the measurement of consumption according to quantitative and qualitative parameters. The reliability of the data can vary widely, not least because in certain areas consumption is invisible to those who would measure it, and so they must make estimates based on assumptions about methodology and sampling techniques. Whereas in auditoria, turnstiles may quite accurately quantify footfall through the premises, and sale or return inspires high levels of confidence in circulation figures for certain types of publication, broadcasters perform to intangible audiences who can be neither counted nor observed en masse. This article considers the appropriateness of sampling techniques used to produce audience research data for the broadcasting industries, for the programmers and advertisers who need detailed 'knowledge' about their audiences. It draws on the contextualization in Cultural Trends 45 (Starkey, 2003), which examined the debate around competing methodologies using either innovatory electronic devices for the measurement of consumption or more traditional human recall. The article raises important questions for those who use sampling techniques in the cultural sector and those who would interpret their data.