2
Political Research Quarterly 00(0)
and beyond the role played
by actual economic perfor-
mance. To do this, we employ a new measure of media
tone developed by applying supervised machine learning
(SML) methods to thousands of newspaper articles in
four national newspapers and validated by human coding.
We make new contributions on the questions of how eco-
nomic performance and media
tone affect collective eco-
nomic attitudes by estimating models of media tone and
economic attitudes that are “saturated” with economic
indicators and then analyzing the relationship between
the residuals from these two models. The approach allows
us to determine whether the portion of economic news
coverage that cannot be explained by economic funda-
mentals has a significant relationship
to the portion of
citizens’ economic attitudes that also cannot be explained
by economic fundamentals. The evidence we present sug-
gests that it is not merely the case that media tone and
mass economic attitudes move together. We find that a
substantial relationship between
media tone and public
opinion persists even after taking into account actual eco-
nomic performance.
Do'stlaringiz bilan baham: