There's this interesting study out of Finland that examines an age-old question -- do the media matter when it comes to what people know? And on top of that, the authors argue the kind of knowledge you measure can make a difference in media effects.
This study looks specifically at newspapers. Exposure (or attention) to newspapers is generally associated with political knowledge. The more you read, the more you know, or at least the better you do on tests scholars come up with to measure political knowledge. The authors look at something they call overall knowledge and something they call structural knowledge. The former is exactly what it sounds like, a lot of different questions used to create an overall index. The latter is based on questions about the structure of government, in this case such questions as the task of the prime minister and some of the principles of the parliamentarian system.
What did they find? Newspaper reading largely disappears as a predictor of political knowledge once you enter a bunch of other socio-demographic-political factors into the model. The authors appear more surprised by this than I am, but let's give them the benefit of the doubt. This lack of an effect really shows up in their structural knowledge dependent variable.
Why? Because they're misthinking this thing. Structural knowledge is best thought of not as a dependent variable, at least not in a news media sense, but as an independent control variable. It would be a great dependent variable if you were studying civics classes, but here it's best thought of as a control factor to see what people know about current events or that sort of thing. Still, it's another piece of the puzzle.