Wednesday's Philadelphia Daily News reported the findings of a new Keystone Poll as follows:
Poll: Obama leads in Pa., but not by much
Obama leads McCain by eight points, 44 percent to 36 percent among registered voters, and by five points, 46 percent to 41 percent, among likely voters.
Poll director Terry Madonna said that the survey shows that Obama hasn't been able to pull away despite signs it should be a big Democratic year in Pennsylvania.
It's utter bullshit. Barack Obama -- based on this poll's own raw data -- actually leads John McCain among registered voters by thirteen and among likely voters by a ten point margin. Let me explain.
The key is sample demographics, and this something on which I've criticized the Keystone Poll for more than three years. The poll, which is run by Franklin & Marshall College in Lancaster, PA, consistently oversamples its Republican part of the state, undersamples pro-Democratic Philadelphia and the Southeast, and produces results that are always skewed against Democrats. When I called once to ask them if they corrected for turnout, the response was "we can't predict turnout."
And that's just false, and lazy. The Keystone Poll has the following regional sample:
I now have a final few questions for statistical purposes only.
CNTY. Region of state (What is the name of the county you live in?)
35% Central
15% Southeast
14% Northeast
11% Allegheny
10% Southwest
8% Northwest
7% Philadelphia
Here, by way of comparison, is the 2004 CNN Exit Poll breakdown of the Pennsylvania vote:
VOTE BY REGION
Philadelphia (12%)
Phila. Suburbs (21%)
Northeastern Pa. (14%)
Pittsburgh Area (23%)
Central/North Tier (30%)
[Based on Keystone's regional definitions, Phila. Suburbs = "Southeast", Pittsburgh Area = "Allegheny" + "Southwest", and "Central/North Tier" = Keystone's "Central" + "Northwest".]
The 2006 PA-SEN and PA-GOV race had almost the same breakdown. Now, you see the problem: Philadelphia and its suburbs should be 33% of the sample instead of 22%, Central and Northwest PA should be at 30%, and instead takes up 43% of the state. That's a pretty big swing between what Keystone polled and how Pennsylvania actually votes.
How big? The math's not hard. (Okay, granted: I'm a former high school mathlete.) If you take the regional breakdown that Keystone provides -- at the bottom of page 9, Table A-1, and multiply the results for each region by the percentages in the 2004 exit poll breakdown, you can extrapolate out a new set of results re-weighted to actual 2004-06 turnout.
This GoogleDoc shows my calculations, and the result is instead of a 44%-36% lead among registered voters, Obama actually leads 47%-34%.
And if that lead expands from 8% to 13%, then it is reasonable to believe that Obama's lead among likely voters is not six percent, but is actually closer to ten percent, about 48% to 38% over McCain. Terry Madonna, please do a better job next time.
One more thing: In 2004, George Bush beat John Kerry 63%-37% in the Central & Northern Tier. If Barack Obama is only trailing McCain by a point or two in these traditionally Republican areas, the Pennsylvania race is over. He's going to come out of Philadelphia alone with a 400,000 vote lead, and there will be no way for McCain to catch up. None.