Regarding Rasmussen's Pennsylvania Poll

Rasmussen came out with a new poll from Pennsylvania earlier today that showed that Obama is only ahead in the state by 4 points there, 51-47. This is a large leap from the 7-point lead that Obama had in a poll that they took two days ago. This struck me as bizarre, so I looked into how the poll was conducted.

Here are some of the internals from the Pennsylvania poll they took today, and the internals from the poll they took three days ago are in parentheses:

Overall:
Obama--51% (53%)
McCain--47% (46%)

Among Whites:
Obama--46% (46%)
McCain--52% (53%)

Among Blacks:
Obama--80% (93%)
McCain--20% (5%)

Notice that the totals among whites remain virtually unchanged. But are we really supposed to believe that McCain jumped 15% among African Americans in three days? No.

These numbers are fishy to say the least. Nothing drastic has changed the race in the last three days, and even if it did, I would expect Obama's support among whites to go down, not among blacks. This is just a sampling error that Rasmussen made, and should be taken into account before all you Democrats start worrying too much.

Now, I have always thought that Pennsylvania was closer than the polls were indicating. I did not believe that Obama had a double digit lead over McCain a week ago. It is apparent that McCain is closing in the state to some extent--the Morning Call tracking poll in Pennsylvania shows that McCain has gained 3 points on Obama in the last two days. They have McCain down by 8 points in the Keystone State.

I believe that the margin is somewhere in the middle of these two polls, and Obama has a 6 or 7-point lead over McCain. But seeing as this where the race is likely to be decided, let's look at some Pennsylvania voting history real quick. In every election since 1984, Pennsylvania has gone bluer than the rest of the country. Take a look.

1984
Pennsylvania--7.3% Republican Spread
National--18.2% Republican Spread

1988
Pennsylvania--2.3% Republican Spread
National--7.7% Republican Spread

1992
Pennsylvania--9.0% Democratic Spread
National--5.3% Democratic Spread

1996
Pennsylvania--9.2% Democratic Spread
National--8.5% Democratic Spread

2000
Pennsylvania--4.2% Democratic Spread
National--0.5% Democratic Spread

2004
Pennsylvania--2.5% Democratic Spread
National--2.4% Republican Spread

Now, let's take a look at the 2008 Pennsylvania and National averages based on recent polling:

2008
Pennsylvania--9.0% Democratic Spread
National--6.2% Democratic Spread

So basically, if Obama wins the popular vote, he will win Pennsylvania. Barack Obama's campaign manager David Axelrod announced today that Obama will not be visiting Pennsylvania in the last three days, which shows that they are fairly confident in winning the state. McCain's internals probably look a lot like the Rasmussen numbers, and that is why they are making this last-minute push.

Also, in 2004, John Kerry overperformed the polls by about 1.5% in Pennsylvania, so that should be taken into account when analyzing these numbers.

So, it still appears as though Obama does have some breathing room and I do think that he will win here. But I wouldn't take the state for granted.