Clinton crushes Obama in KY, but her PA lead shrinks in 2 polls

Many of the upcoming primary states have never been polled, so Survey USA's decision to release a poll from Kentucky last night was particularly exciting:

• It shows Hillary Clinton crushing Barack Obama 58% to 29%. Obama only gets 23% of the white vote, and is trounced by 37% among women.
  

Clearly, put Kentucky in with the states Clinton is favored to win -- but the primary is not until May 17th, the same day as Oregon. It is unclear, therefore, if Clinton will ever have a chance of winning Kentucky, or whether the results in Pennsylvania, Indiana and North Carolina will force her out before May 17th.

The latest Pennsylvania poll from Rasmussen is not reassuring for the Clinton campaign, as it shows Hillary's number falling dangerously quickly and getting to what I believe is the smallest margin we have seen yet in any Pennsylvania poll:

• Clinton is up 47% to 42%; she was up 10% last week and 13% two weeks ago.

Obama's fundraising edge is allowing him to outspend Clinton on Pennsylvania television, which will help him keep the gap close. But the Rasmussen poll clearly identifies another factor in Hillary's decline and notes that the Wright controversy has been replaced with another scandal, this one hurting Clinton, as the story on Bosnia misstatements seems to have touched a nerve: 47% of voters say they have followed the story "very closely," and 27% said "somewhat closely."

We will naturally need confirmation of this poll before concluding that Clinton is in danger in Pennsylvania. The last non-Rasmussen poll still showed her up in double-digits late last week. But keep in mind that it is not just a defeat here that would pretty much end her run: Clinton needs to score a major coup (as big if not bigger than in Ohio) to generate momentum and produce doubts about Obama's capacity to close it off. She particularly needs to trounce him among blue-collar voters to have more support for her argument that Obama cannot win that group. A 5% victory will not get that job done, at all.

Meanwhile, a very interesting polling story form Pollster.com's Mark Blumenthal, who produced an analysis of the Rasmussen and Gallup tracking poll patterns, explaining that Obama consistently goes up in the weekdays in the former and in the weekends in the later. In other words, Blumenthal concludes, the polling methodology of both groups makes the results dependent on the days of the week the polls are taken in (which is a somewhat well known phenomenon, since the sociology of people who pick up the phone changes on Saturday or Sunday).

The conclusion of this is that attempts to use the Gallup and Rasmussen polls to draw daily conclusions about the evolution of the race is not appropriate, contrary to the way they are typically being used. (Note that I rarely reported these tracking polls anyhow, finding them only interesting when broad trends were being revealed, such as Obama's significant 10-day decline after Wright).

Update: We did not have to wait for long to get another Pennsylvania poll, this one from Survey USA:

• Hillary Clinton is leading 53% to 41%, a clearly more satisfying lead than the one she has in Rasmussen.
  
• But this poll too shows Obama picking up a lot of ground from SUSA's previous poll released 3 weeks ago; he then trailed by 19%.

This complicates the picture since Clinton has a satisfying lead but is declining. Interestingly, the female vote stayed almost exactly the same and it is the switch of the male vote that allowed Obama to pick up ground. Also, the white vote did not move (Clinton gets 62%) but Obama extended his margin among blacks by 12%.