Interesting column online by the Wall Street Journal's "Numbers Guy" that looks at the Pew research on what media people say they use and their political knowledge. It's more about what media people say they consume than discussion of political knowledge, but still kinda interesting, especially the stuff on social desirability when asked what news or television we watch.
We know social desirability is a problem in survey responses. For example, we underestimate how much TV we watch and when reporting our TV viewing, we overestimate how much "quality" or "highbrow" stuff we watch (PBS) and underestimate how much crap we watch (Dancing with American Idol's Survivor).
It's an interesting problem because overestimation of news viewing could scram the numbers, create artificial relationships with measures of political knowledge. Higher or lower, I'm not sure. Gotta think it through, and to do this you'd have to consider what kind of people are more likely to overestimate news consumption.
There are measures of this, by the way. Sometimes in surveys we'll try to get a sense of someones likelihood to answer questions in a socially desirable way and then control, statistically, for this possibility. But you don't see this done very often in surveys. That's probably too bad.