My projection had Clinton winning 60 percent of the two-candidate vote in Pennsylvania. But in the RealClearPolitics.com average of recent Pennsylvania polls, she's getting only 54 percent of the two-candidate vote. Obama is clearly leading in Philadelphia, where blacks will make up about half of the electorate; Quinnipiac has Obama ahead in the Philadelphia suburbs, but SurveyUSA has metro Philadelphia even, which must mean that Clinton is ahead in the suburbs.
My projection had Obama winning 55 percent of the two-candidate vote in North Carolina. Current polling gives Obama 61 percent of the two-candidate vote there. Curiously, the share of undecideds in North Carolina averages 20 percent, compared with 10 percent in Pennsylvania. Perhaps this just reflects the much heavier campaigning in Pennsylvania in recent weeks.
In both Pennsylvania and North Carolina, we have seen the same kinds of surges of support to Obama in periods when he was heavily outspending Clinton on television as we saw in similar periods in Ohio and Texas. In Texas, Obama seemed to seize the lead, while in Ohio, Clinton seemed to maintain a small lead. In both states, Clinton seemed to get an uptick just before the primary, and she ended up with 55 percent of the two-candidate vote in Ohio and 52 percent of the two-candidate vote in Texas.
Can she do the same in Pennsylvania? I think she needs to. If she gets only 54 percent of the two-candidate vote in Pennsylvania April 22 and if polling shows Obama running away with North Carolina and being competitive in Indiana, there could be a cascade of superdelegates toward Obama. Even if there isn't, a similar phenomenon could occur if Obama wins North Carolina by anything like his current margins in the polls and especially if he wins in Indiana as well. I still think Clinton is set to carry Kentucky and West Virginia by wide margins. And Puerto Rico, with 2.5 million voters, 80 percent of whom turn out in Puerto Rico elections, remains a huge question mark. But will she get there?
By Michael Barone