Here is a well-known but now somewhat dated (1991) paper by Zaller and Chiu. It suggests two regularities:
1. U.S. press coverage tends to take its positions from the range of views which exist within government (“indexing”).
2. When a foreign conflict goes well, the U.S. press becomes more hawkish; when the conflict goes less well, the press becomes more dovish. The press swing in opinion is stronger than the swing of opinion from official sources.
Here is an empirical paper, applying this framework to the Libya crisis of 1985-1986. Here is a general look at the indexing hypothesis, again dated and pre-blogosphere. Here is a 2008 paper, showing greater influence for media, relative to the distribution of opinion within government.