Real Climate has posts here and here on the function of press releases on the dissemination of scientific information.
A couple things to note. First many journals, including Science and Nature send out press releases for major articles they publish. (This was something I missed on my scientific communications chart, which only talked about university press releases.) As Real Climate notes, these press releases are written without consulting the scientists, so that scientists contacted by journalists for a quote often have to ask the journalist what the press release says they said. Science and Nature also both have embargos on the publication of a paper before it appears in their pages, increasing the importance of the press release they put out.
For their first field project my scientific reasoning class had to analyze a report of a scientific event from the popular media. Interestingly, four of them chose this article on decreased polar ice coverage. (Here is my answer sheet. As usual, people confuse the real world with the model and the model with the prediction generated by the model.) Tomorrow they are supposed to bring in the primary source for that article as a part of their second field project. This is the source for the polar ice article, but I doubt many students will have found it. It has not been published yet, and when it is published, it will be in a journal that my university does not subscribe to, and I imagine few do. I only found it because there was a small link to it from the New Scientist article on this research. The primary source is hosted on NASA's servers, and linked to from some of their pr pages, but I don't think any students will find their way there.
I think I'm going to add one more step to their second field project. I want them to find the press release that the newspaper articles were drawing on, and compare the primary source, the press release, and the newspaper article. This should be fun.