As part of the Advancing Understanding of Media Effects on Political Knowledge project, we conducted a laboratory based experimental study to investigate some of the issues we identified in our secondary data analysis. It became obvious in our analysis that the measurement of exposure to media was critical and has been measured in many different ways. Therefore we tested various measures in an experimental study.In this project we use three publicly available secondary survey data sets--the 2009 European Election Study, 2006, 2008, and 2010 European Social Survey, and 2005-2010 British Election Study Panel and a lab study (source of archived data)--to examine 1) the substantive issue of the relationship between media coverage and political knowledge and engagement, and 2) four different methods to gauge media effects. Our focus is on the traditional media of television news and newspapers; while social media have a role to play their influence is still dwarfed by traditional media. We will link media content data that we have or that are available to each of these surveys in order to estimate the amount of exposure to political information, its subject matter, tone etc. and the consequences for political attitudes and behaviour. These data sets are cross-national, allowing us to examine the interaction of different media systems, characteristics of media content, and individuals. Our project will show the potential of secondary data for answering fundamental questions about European democracies and the effects of media, particularly timely in an era that spans the worst financial crisis since the Great Depression of the 1930s when there are questions about people's continuing faith in democracy, and when the role of media in contemporary society is under scrutiny through bodies such as the Leveson Inquiry; while also answering the important practical questions of how to apply different methods of estimating media effects, when they are appropriate, the kinds of relationships they uncover, and when and why their findings conflict.

Experiment with follow-up survey. The proposed experiment would employ an experimental benchmark, in which randomly assigned subjects were exposed to “media coverage” in a controlled setting. We compared the self-reports of subjects who had previously been assigned to treatment with the self-reports of those assigned to control. The advantage of an experimental benchmark is that the researcher knows who has been assigned to receive treatment. It also means individuals cannot self-select into the treatment. Other (non-experimental) benchmarks may still have error—e.g., misreporting with the “people meters” used to construct Neilsen ratings or social desirability—and individuals choose their own level of exposure.