Just a quick thought about how the "closed" nature of tomorrow's primary might affect the outcome. The conventional wisdom is that it hurts Obama, since he tends to do better among independents and Republicans. But I think that could be wrong. My sense is that Hillary and Obama are basically running even among the independents and Republicans--call them I&Rs--who'd vote in the Pennsylvania Democratic primary if it were open. (I base that on Ohio, a demographically similar state with a basically open primary in which I&Rs split down the middle. Whether that was a function of strategic voting, or an increasing affinity for Hillary, I can't say.)
But I also suspect that Obama's I&Rs support him more intensely than Hillary's. Why does this matter? Because in order for an I or R to vote tomorrow, they'd have to have registered as a Democrat as of March 24. As a general rule, only relatively intense supporters would go through the trouble of doing that, which, as a general rule, means Obama I&Rs. My guess is that if you polled only the I&Rs among the 325,000 Pennsylvanians who became Democrats over the last year (or the 200,000 who did so since January 1), you'd find Obama with a pretty big advantage. If, by contrast, you polled all the I&Rs who would have voted if the primary were open but won't because its closed, my guess is that they'd favor Hillary by a wide margin. In a nutshell, the closed primary discourages the Hillary's I&Rs disproportionately.
P.S. For what it's worth, the same AP piece I linked to above cites a Franklin and Marshall College poll in which 62 percent of newly-minted Pennsylvania Democrats said they favor Obama. That's not quite the same as what I'm talking about--it includes previously unregistered voters--but you it gives you a rough idea.
Update: TNR alum Marin Cogan had a nice piece today breaking down the demographics (and likely preferences) of all those newly-registered Democrats. Marin comes up with yet another registration figure: 306,918 since January 1. That could be, since the 200,000 figure I cited above came from an item written the day of the deadline, by which point all the new voters probably hadn't been counted.
--Noam Scheiber