pretty good. the title is a bit awkward, but I know what youre trying to
do. the rest i think would cause me to read the paper, but be suspecious
that you can determine causal direction from these data, but that's
presumably also a problem in the original paper. your comment about ols
supressing information is not clear; perhaps you can emphasize what it is
about negbin models that reveals new information, and then what precisely
what that info is.
Gary
On Sun, 30 Apr 2006, Jacqueline Chattopadhyay wrote:
Hi all,
Please find our preliminary abstract below. We would appreciate comments and
recommendations. Thanks,
Jacqueline and Jen
The Media Followed the Parties, but Not Equally on All Issues: A Closer
Look at Agenda Formation in the 1997 British General Election
A small body of literature has attempted to determine the causal flow of
agenda-setting during political campaigns: do the media lead the parties or
vice versa? Using time-series cross-sectional data from the 1997 British
general election campaign, Brandenburg (2002) employs ordinary least squares
regression to conclude that the media tracked the parties' issue agendas. We
replicate these results but find evidence in first differences that parties
conceivably followed the media agendas as well. We therefore re-estimate
Brandenburg's five models using negative binomial regression in light of the
over-dispersed count-data. The results confirm Brandenburg's stated but
imperfectly supported conclusion that agenda formation was unidirectional,
from parties to media. In addition, however, the negative binomial models
uncover differences across policy dimensions in the magnitude of this party
impact, information that OLS suppresses. It appears that the parties
particularly affected the media's tendency to cover the economy, welfare,
education, and foreign policy.