Measuring the Media Agenda

Abstract

Measuring media attention to politically relevant topics is of interest to a broad array of political science and communications scholars. We provide a practical guide for the construction, validation, and evaluation of time series measures of media attention. We review the extant literature on the coherence of the media agenda, which provides evidence in support of and evidence against the emergence of a single, national news agenda. Drawing expectations from this literature, we show the conditions under which a single national news agenda is likely to be present and where it is likely to be absent. We create 90 different keyword searches covering a wide range of topics and gather counts of stories per month from 12 national and regional media sources with data going back to 1980 where possible. We show using factor analysis wide variance in the strength of the first factor. We then estimate a regression model to predict this value. The results show the conditions under which any national source will produce time series results consistent with any other. Key independent variables are the average number of stories, the variance in stories per month, and the presence of any “spike” in the data series. Our large-scale empirical assessment should provide guidance to scholars assessing the quality of time series data on media coverage of issues.