In one study by Robert Vallone, Lee Ross and Mark Lepper, pro-Palestinian students and pro-Israeli students at Stanford University were shown the same news filmstrips pertaining to the then-recent Sabra and Shatila massacre of Palestinian refugees by Christian Lebanese militia fighters in Beirut during the Lebanese Civil War. On a number of objective measures, both sides found that these identical news clips were slanted in favor of the other side. Pro-Israeli students reported seeing more anti-Israel references and fewer favorable references to Israel in the news report and pro-Palestinian students reported seeing more anti-Palestinian references, and so on. Both sides said a neutral observer would have a more negative view of their side from viewing the clips, and that the media would have excused the other side where it blamed their side.
It is important to note that the two sides were not asked questions about subjective generalizations about the media coverage as a whole, such as what might be expressed as “I thought that the news has been generally biased against this side of the issue.” Instead, when viewing identical news clips, subjects differed along partisan lines on simple, objective criteria such as the number of references to a given subject. The research suggests the hostile media effect is not just a difference of opinion but a difference of perception.
This effect is interesting to psychologists because it appears to be a reversal of the otherwise pervasive effects of confirmation bias: in this area, people seem to pay more attention to information that contradicts rather than supports their pre-existing views. This is an example of disconfirmation bias.
Wikipedia article with related subjects and suggested follow up reading.