I would say I'm a Strong Contextualist when thinking about how we act in our different social roles. In particular, I don't think there's much evidence of carryover from how we act in our online persona and our meatspace performances.
In fact, I don't think there's always a lot of carryover in our meatspace performances either -- I think folks can act pretty differently in their birth family, their own adult family, their classrooms, faculty meetings, interactions with Deans, donors, and university staff, and so on. And they can act differently at different times in those roles: you can perform well in some faculty meetings and not in others, and we have to avoid hasty generalizations and judging by the worst in thinking about the overall performance in that role.
Given this Strong Contextualism, I think it's very dangerous to try to infer anything about performance in any one role from performance in another, and especially when crossing the online / meatspace divide. The dangers here are evident in the anti-Salaita camp's attempt to infer possible classroom behavior from his tweets (all the while ignoring the evidence there is about his classroom behavior).
And I think it's less dangerous, but still inimical to a vigorous professional debate culture [edit: but *should* we have that?], to think that criticism of a particular performance is criticism of the person. Thus attacks on what someone said at time T1 are not attacks on that person, and so one should not appeal to what they do at home or in the classroom or in the department or in the Faculty Senate in order to mitigate criticism of behavior online or in another meatspace sphere. "But he's a devoted son to his elderly mother" is not really a good response to "man, you really botched could have done better with that argument." [Edited, in response to seeing this.] Doing so is just the flip side of the bad inference from online behavior to meatspace behavior.
Of course, criticizing someone's behavior in any role should also avoid hasty generalization and judgement by the worst. That can be avoided by demanding that criticism of behavior in any sphere follow the rules of scholarly criticism: make accurate quotes and establish the context, including the rules of the game of the genre involved.