I often hear liberals complain that pollsters don’t include cells phones (or enough cell phones) in their sampling, and that they are therefore skewed against Democrats. One obvious argument in support of this theory is that younger people tend to favor the Democrats and they tend not to have land-line phones. Nate Silver takes a look at the pollsters and finds a 2.8% bias against Obama for polls that do not include at least some cell phones in their samples. Other recent studies have found smaller biases or none at all.
Perhaps the most interesting finding in Nate’s analysis is that Gallup polls over the course of the year have averaged out to a 0.6% bias toward McCain over the mean. It is the only pollster that includes cell phones to show a bias in McCain’s favor. That makes their finding today all the more significant: Obama 50, McCain 44. That matches up well with today’s Research2000 poll: Obama 50, McCain 42.
If the assumption is correct that McCain will pull in about 70% of undecided voters, the result would be Obama 52, McCain 48, or Obama 53, McCain 47. Throw in some third party voters in there for shits and giggles, but you get the idea. However, these national polls can be deceptive. In the Research2000 poll, we see the following regional results:
McCain/Obama
NORTHEAST 33/60
SOUTH 52/39
MIDWEST 38/53
WEST 40/52GEOGRAPHIC BREAKDOWN:
Northeast:
DC, ME, VT, NY, MD, PA, CT, DE, MA, NH, RI, WV, NJSouth:
FL, NC, SC, AL, MS, GA, VA, TN, KY, LA, ARMidwest:
IL, MN, MI, OH, WI, IA, MO, KS, IN, ND, SD, OK, TX, NE, COWest:
NM, CA, OR, WA, AK, HI, MT, ID, UT, NV, AZ, WY
When you factor in that the Midwest number includes Texas, Oklahoma, Kansas, Nebraska, and the Dakotas, Obama’s 15-point lead looks even bigger. This is especially true because, other than Illinois, it is not expected that Obama will win any of the Midwest states by huge margins. For this reason, there is very little in the Blue States to offset McCain’s big numbers in places like Oklahoma (where McCain leads 63-36) or Kansas (where McCain leads 59-40). If Obama has a 15-point lead in the Midwest despite his pitiful performance in the Plains states, then he is truly sitting pretty.
Meanwhile, McCain leads only in the South. And McCain must carry every southern state, including Virginia and Florida (where Biden and Obama are campaigning today) in order to have any chance at victory. The polls may tighten up again, as they have in the past. But the national polls are deceptively close because of the regional differences.
National polls are virtually meaningless – but so are regional polls. State by state polls (and congressional districts, in Nebraska), and adding the electoral totals, are all that matter. And while McCain basically pulled even at the height of Palinmania, Obama is pulling back ahead in the electoral count and his long-term trends look good. Fragile, to be sure, but encouraging.
Nothing else counts.
Another poll today from Rasmussen, which tends to have a bit of a house effect favoring McCain, actually has South Carolina now within the MoE.
McCain’s above 50% in it, but, still, pretty impressive.
The polls at the state level are starting to back up the nationals from Gallup and the Great Orange Satan.
How many minority and college voters will be disenfranchised in these battleground states, I wonder?
It won’t take many. Figure out how many it’ll take to give McSame a 51%-48% win in Ohio, Pennsylvania, Michigan and Colorado.
Game over.
538 projection [graphic representation] from monday:
today:
that’s a significant change, given their methodology.
and if florida’s so important, why’s palin canceling her appearances there? things in camp mcstain-o-palin appear to be in a state of turmoil, eh.
30% chance of rain…in florida in sept…whodathunk.
Maybe because every time she speaks, they lose votes.
She is only running Palin 1.0, is only able to regurgitate a limited set of right-wing talking points, and is getting to be regarded as not very stable.
Palin 2.0 was supposed to be better, but the release got delayed past election day.
this unvetted selection has turned on mcstain and the RATpug’s with a vengeance.
the old expression pig in a poke comes to mind.
couldn’t happen to a more deserving fellow than j-mac and his enablers…bwahahahahahaha!
A pig with lipstick!
Tell me you’re NOT calling the Saintly Miss Sarah a PIG! You WOULDN’T!
Nooooo! Not me!
Dada, there was a 30 percent chance of rain in Florida so she canceled. The Hockey mom can not get wet! This excuse is so pathetic that it is hilarious.
…ALL those people who screen their calls.
Everybody I know. ALL of my colleagues and co-workers and friends.
Busy professionals do not sit next to their telephone waiting for it to ring with an offer for carpet cleaning or tickets to a police fundraiser or random survey-taker. If it is somebody we know, we pick it up. Otherwise, it gets ignored. I clear out my message machine of detritus once a month or so.
Since the automated dialers do not press *62 and do not leave messages and do not wait for the beep, or any of the other shields, they simply do not reach the people who are entrepreneurs, contract-workers, or busy managers.
My Grandma would wait by the phone for gossip… but she was almost 90 and handicapped when she died. So the pollsters reach old folks and stay-at-home mothers. Whoopi-dink. THAT is a sampling?!?
They are beginning to sample a few cell-phones. That is a start. Perhaps they’ll get some figures for what percentage of the electorate is cell-phone only, and can adjust their statistics.
Maybe in another few years, they’ll find a way to reach the professional class. Or be able to determine what percentage of the electorate they are still missing. Maybe.
Meanwhile, anybody who votes for a guy because some pollster tells them “this guy is a winner” is as stupid as an investor who buys a stock because it is touted on a message board.
Sheep move in a herd. When sheep get fleeced, is the wool pulled over their eyes?
Booman, I’m glad that you’ve diaried this, but we need to keep the issue alive… lest people assume that the talking heads are waving accurate numbers or maps, or that the numbers have any relationship with reality.