As our lives have become more digital, so has the recognition that political participation can have both offline (traditional) and online aspects.
The traditional measure of participation is giving money, putting a campaign sign in your yard, attending a rally, the sort of "offline" activities that we normally think of when we say we participate in politics. Online is different, from "liking" on Facebook to "retweeting" via Twitter to almost anything we can manage to do with our opposable thumbs and a keyboard.
This 2012 study (in the Journal of Computer-Mediated Communication) looks at both types of participation forms and includes, in Table 2, a number of factors thought to predict them -- including political knowledge and news exposure. It also includes some structural equation modeling, which for the statistically uninitiated can be a fearful thing. My advice -- just admire the pretty pictures and don't worry about the smoke-and-mirrors underneath the analysis strategy.
It's interesting that political knowledge predicts offline (traditional) but not online participation. The roles of social media, they look largely the same for both forms of participation, though the generic "media use" factor is certainly a much stronger predictor of offline than online participation. That makes sense, given "media use" is more traditional, thus it predicts the more traditional of two participation measures.