Pennsylvania New Democratic Registration Map

This map is based off of the last three weeks' worth of registration information in the great state of Pennsylvania. What this shows is the number of new Democrats—either party-switchers or new voters—per 1,000 Bush voters in 2004. This gives us a sense of how fast a county's Republican or apathetic electorate is moving away from the GOP. In order, the biggest movement has been in the following regions:

• Philadelphia
• Harrisburg
• The Philadelphia Suburbs and orbiting towns—Bucks, Montgomery, Delaware, and Chester County, which have been trending blue recently, but also now Berks County (Reading), which is significantly less affluent.
• Northeastern Pennsylvania—Scarnton, Wilkes-Barre, and two more rural counties to their south.
• State College
• Erie
• Lehigh Valley—Bethlehem, Allentown, Easton, etc.

The registration picture appears to favor Barack Obama; more people live in Philly, its suburbs, and State College than SWB, Lehigh Valley, and Erie. But there are two caveats. First, the advantage is very small, perhaps 3% of the primary electorate at most. And Second, the affluent suburban Philly voters might not go for Obama; he won those voters in Texas, but not Ohio.

Comments

[John]

Other than the new registrations in Scranton-Wilkes-Barre and Erie, I think all these regions help Obama. New voters have generally been strong for Obama, and areas like Harrisburg and the Lehigh Valley are meant to be swing regions, not Clinton strongholds. New voters there (to say nothing of new voters in Philly, its suburbs, and State College) ought to favor Obama.

[Nicholas Beaudrot]

I just looked at the census figures, and Dauphin county (Harrisburg), with high registration, is 18% African-American, while neighboring Cumberland, with low registration but similar population, has only 3% African-American. The other thing that may be a factor there may be the Rendell machine. Hard to tell.

Still not sure about Lehigh. Demographically it looks like the state as a whole, with slightly higher income.

And either way, we're talking about at most 4-6% of the primary electorate.

[Joe]

"primary electorate" = registered Dems or projected # of Dem participants?

[lutton]

Berks Ct (Reading) provides a base for Rep. Jim Gerlach to continue to win a seat in Congress. If Berks is trending away from the GOP, that seat becomes easier to win.

[lutton]

And wow, Lancaster Cty sticks out like a pale blue thumb among the leftward-trending greater southeastern region.

Is it really that conservative (the last bastion in the Philadelphia region!) or does it somehow relate to the Amish population?

[Nicholas Beaudrot]

Projected # of Dem participants. But I may have been highballing turnout based on open or semi-open primaries. The closed primaries in CT and NY had about 66K and 62K voters per CD. Delaware had 91K voters in one CD, but it's a very large CD. Louisiana had 51K, but it's lost a lot of population. Maryland had 82K or so. So, let's say 70K voters per CD; we'd be at 1.33M voters statewide. If Obama wins the new voters 70-30, which is probably the upper bound, the new voters would take a 55-45 win to a 52.2-47.8.

Now, this time, unlike Super Tuesday, the campaigns only have operations in three or four states. So they may be able to juice turnout a little more. If turnout is 2M, which is probably the upper bound, the new voters would net Obama just under four points, from 55-45 to 53.1-46.9. And I don't know if these voters are being polled or not ... they probably are.

[John]

Only somewhat related, but I just did a little math with the delegate counts. Basically, I assumed the floor delegates for each candidate, and then the delegates which would swing depending on who won. I assumed an overall popular vote floor of 40% for Obama and 47% for Clinton, and then various floors for each in each of the congressional districts.

Absolute best case for Obama from the state as a whole would appear to be +11. Absolute best case for Clinton would be +32 - i.e., a 43 delegate swing. 16 of that swing comes from statewide totals. The rest comes from the various individual districts. Given that, Bowers's estimate of Clinton +10 appears to be right in the middle of the range of possibilities.

[Nicholas Beaudrot]

lutton: I think the weak vote in Lancaster may be because Obama didn't have a field office there until last week. When the DoS posts the final statistics we'll have a better picture.

[Hal]

I'm not sure why projected new Democrats per 1,000 Bush voters in 2004 is a useful measure. For example, Philadelphia didn't have a lot of Bush voters in 2004, so a large percentage swing of a small group may not amount to much on April 22. Am I missing something?

[Hal]

The 2004 vote in Philadelphia was:

Kerry 542,205 81%

Bush 130,099 19%

[Joyful Alternative]

Anecdotal newspaper reports had Obama people bringing in 10 times the volume of new registrations to county offices as Clinton people.

[Nicholas Beaudrot]

Hal, believe it or not, if you just do new registrants per 1,000 voters, Philadelphia is still 12th.

JA: I can believe those numbers, but if they're just registering people at bus stops there's no way they can be registering only Obama voters.

[Nicholas Beaudrot]

Right ... to answer the broader question, new voters per 1,000 Kerry votes gives you the impression that the Democratic party is growing by leaps and bounds in Appalachia. But, a 10% increase in party size, when the party is only 25% of the public, gives you a party that's 27.5% of thet public. A 3% increase in party that's 80% of the public gives you one that's 82.4% of the public.

The "per 1,000 Bush votes" gives you a measure of "how fast are voters deserting the Republican party". I suppose I could just look at switches, which would be more informative on that front. It does seem that African-Americans are underregistered.

[shoez]

post something new jerkface

[Northern Phila. suburbs]

if they're just registering people at bus stops there's no way they can be registering only Obama voters.

Don't forget, in PA you use the same "new registration form" regardless of whether you are re-registering with a change of party, change of address, change of name, etc.

So, yes, no need to assume that every person in a bus station who stops to fill out a form is an Obama voter, but Clinton/McCain voters aren't usually going to be among the change-of-party folks, so that cuts down the numbers somewhat. Plus, I suspect there is some self-selection going on. Although that is based on anecdote, I am right here at Ground Zero for the next few weeks.