Here's an interesting field experiment in which authors randomly assigned real live people to receive either the Washington Post or the Washington Times or to a control group of neither newspaper to see what effects were found. I love real-world studies like this. Sure, they're messy, but they're also generalizable beyond the intensive study of the college sophomore.
Okay, enough bashing. It mattered not at all whether subjects received the Post or the Times on political knowledge, turnout, or attitudes. Receiving a paper did result in more support for the Democratic candidate (interesting given the Times is considered conservative). The authors try, not very satisfactorily in my opinion, to explain this effect. I don't envy them on this. It's hard to figure.
As an aside, it is kinda odd to find this published American Economic Journal. But, in defense, I've seen a few bias/slant studies of media content in economic journals, so maybe this makes more sense that it seems on the surface.