Morning Jay: California Polling, Castle Bows Out, and the Politics of the Census!
6:30 AM, Sep 30, 2010 • By JAY COST
Republican Senate candidate Carly Fiorina
That’s by about the same amount that John Kerry beat George W. Bush in 2004. Yet Kerry defeated Bush among independent voters by a wide margin, 59-36. In the CNN/Time poll, it’s Fiorina who has a lead over Boxer among independents, and a sizeable one at that, 53-39. The poll also finds the two parties very well sorted, Democrats going for Boxer 93-5 and Republicans going for Fiorina 92-4.
So, the only way to find a nine-point lead for Boxer is if the poll has a huge sample of Democrats.
Unfortunately, the CNN/Time poll does not give the Democrat-Republican-independent spread, but there’s another way to show the poll is sampling too many Democrats. If we take the performance of Fiorina and Boxer among Democrats, independents, and Republicans, then use the exit poll data to get the party spreads for 2008, 2006, and 2004 – we should get a sense of just how big the Democratic overestimate is.
That’s what the following chart does:
What’s the implication of this? It’s simple: The CNN/Time sample is more favorable to the Democrats than even what happened in 2008. If the partisan mix is recalculated to reflect the spreads in 2006 or 2004, then we have a dead heat in California.
Ditto the governor’s race, where Jerry Brown has a huge lead due in no small part to a more pro-Democratic electorate than what even happened in 2008.
Again, the story is the same. The two parties are well sorted, and Whitman has a big lead among independents. The only way to get a nine-point Democratic lead is to sample a more Democratic electorate than even 2008.
I appreciate the need for leeway in partisan sampling, but it seems quite unlikely that the 2010 midterm electorate will look like this.
2. Castle Won’t Run As Write-In Candidate. Word came late last night from the Hill: