So basically, they're doing well with the uneducated/ignorant
I don't entirely agree. He's seemed fairly pro-Obama in the past few days, but I think he's been trying to look for good news for McCain in the polls. He says that. He's just not finding much.
Second STPA on Nate Silver. He's digging for some good news for McCain, and is quick to acknowledge when national polls show movement.
The internal campaign memo is bullshit anyway. By what metric is the race "essentially tied"? Only a few polls show this a 2- or 3-pt. race. That's why Silver aggregates and weights the pollsters based on past performance. It cuts this cherry-picking that campaigns are wont to do off at the knees.
i was just reading in US News and a few other places that Dems are turning out unlikely voters 2:1 in all the swing states.
But, hey, believe the Red Menace Fearfest from the Socialist Red Bushies if you want to.
If you want to embrace your OCD, check out this video about the MSNBC tracking tool they will unleash election day? Want real-time results by precinct on a nation-wide bases? Customized reporting? They've got you covered:
If you don't trust Nate Silver, look at the maps on Real Clear Politics, CNN, or anywhere else and they all say pretty much the same thing. On the NY Times map, Obama's already winning even without the tossups.
@2 - I'd agree. At this late date, both sides are looking for stuff to be optimistic about. It's true there has been a slight tightening in some of the national tracking polls, but Obama retains a solid lead (> 5%) in more than enough battleground states to pull this off comfortably. The Corner is interested in spreading rumors that the race is really tighter than it is, in the hopes of keeping Republican voters from becoming discouraged and not turning out, partially to boost downticket races as well as vote for McCain.
But it's not over 'till all the votes are cast, and that means we can't let up. The bigger our win is, the better. And there's still a chance we could lose, so it's time to pull out all the stops and donate again to Obama/Gregoire/Burner/Prop 8.
TPM and HuffPo are the ones who put up a banner headline every time there is good news for Obama in the polls. FiveThirtyEight reports the same set of polls day in and day out, no matter what they say. He's also adjusting his model every day as the election approaches to give heavier weight to the more recent polls, which is the opposite of what you would do if you wanted Obama to look better. The more recent polls will be tighter than the older ones.
All of the data Silver uses, and the entire mechanism behind his model, are published on his blog. It is completely transparent.
Nate Silver has also written on his blog and in the NY Post about sifting through the figures trying to conjure up some scenario, no matter how far fetched, where McCain could possibly win. Which is, again, the opposite of what you're accusing him of.
Your own Dr. Science was the one squinting at the graphs and trying to project even more rosy scenarios than Silver offers. Funny thing, that.
Check out a less obviously biased poll site like RCP and know hope.
I hope Playboy never does a "Women of Walmart" spread.
I call BS on this story - it's an "internal memo", so we know it's biased, right? They're not going to just tell their volunteers that it's a lost cause. Secondly - there's not _one_ citation in this thing. Where are these numbers coming from? A few minutes with search and replace and you'd have a Kusinitch lock.
Silver seems to be a great statistician, but it feels as though he wades into the figures looking for good news. If he was a scientist going into things looking for a certain result, he’d be doing bad science, right? Does the same thing apply for electoral projections?
And it's an insightful point, Paul. But, your concern is best directed to skepticism of Nate's discussion rather than throwing out his raw data. I.e. don't read the text he wraps around his tables; just look at the tables themselves. Also keep an eye on his methodological changes, that can bias the tables.
@6 I posted this earlier, but you are wrong. O'Reilly has McCain up.
Keep in mind, Fox can be trusted as they are Fair AND Balanced. No one else makes that claim.
McCain up by 6 electoral points.
Is it just me, or does rumors of tightening only help us by not encouraging complacency?
Campaigns do have polling not available to the media or RCP or 538, but chances are those polls show nothing positive for McCain. That memo is most likely a fake, distributed to friendly media by the campaign so they have a story peg to claim that the election is close -- basically, it's a GOTV effort aimed at their base.
Right now, the GOP isn't really trying to win the White House. They're going through the motions, trying to keep the base excited about voting so they show up and vote (R) downticket, and maybe keep the Dems from getting 60 in the Senate.
@12 - Exactly. The middle of fivethirtyeight.com is opinion, the sidebars are math. Silver himself is clearly biased, but there's no reason to think the math is.
If he was a scientist going into things looking for a certain result, he’d be doing bad science, right? Does the same thing apply for electoral projections?
Good scientists do go into things looking for a certain result (it's called a "hypothesis"). But they also define the rules before embarking on the study, and commit to playing by the rules whether their hypothesis turns out to be right or wrong. This is a critical point. Lots of bad "science" (like the industry studies that show tobacco smoke is excellent for your health) are created by manipulating the rules after the data is in hand, so they can produce the desired result.
On this point, I think Nate Silver passes the scientific test. He defines his rules clearly, and checks periodically to see whether the rules match up with reality. Yes, he's partisan. But that doesn't mean he's wrong.
if walmart can be bankrupted before the next election, there will be no 'walmart women'. i can dream, can't i?
It was actually following cochie's (13) link that convinced me things were still going well. O'riley's analysis puts states where Obama has a double digit lead in the undecided category.
@10 - I can't tell if your just being sarcastic, but Playboy has already done a women of Walmart spread.
"Internal McCain memo" should've been your first clue that the projections contained in it are biased as sin. When the ship is sinking, the only way to keep the rats on board is through rosy, if factually-deficient, predictions. This whole thing reeks of desperation.
The primary storyline the GOP needs to advance right now is that somehow the game has changed. All their outlets are running with this. My guess is they'll pin it on Obama's ad-time tonight, saying it somehow swung "Wal-Mart Women" against him or something similarly reasonable-sounding.
These are the kind of cover stories necessary to explain shenanigans that produce unlikely results. Unlikely as in not legitimate.
I disagree completely. Can you site a single aspect of Nate Silver's numbers that imply he's cherry-picking? He has never hid that he's pro-Obama, but statistics speak for themselves -- especially when you disclose your methods as well as Silver does.
I suppose we'll see how correct his model is when the election rolls around, but their is nothing in his method that leans democratic that i can see, and if your going to bash him, at least cite a single thing in his model you disagree with.
You know that the McCain camp knew they'd lost when they started up the desperate and pathetic "let's prove reality wrong"/"Obama's already measuring the Oval Office for drapes" rallying cry in speeches that worked so well for Bush/Quayle and Dole/Kemp
That's some hilarious spin at the O'Reilly factor...
Oregon, Wisconsin, Michigan, Minnesota and Iowa where McCain has given up and Obama is ahead by double-digits are 'swing states'.
Indiana, West Virginia, North Carolina, Georgia, North Dakota, Montana and Arizona where polls show Obama either leading or only down by single digits are not.
The Fox site is hilarious -- WI, MI, MN, OR, and IA are undecided! IN and NC are decided for McCain! Right. Sure they are.
Nate Silver has never hidden the fact that he is an Obama supporter. But that doesn't mean his numbers are wrong. Math is math. His modeling methods are completely transparent. I do read bias in his commentary at times, but I see no bias in his math.
But lets say for the sake of argument that he's totally in the tank for Obama, and is cherry picking the polls and rigging his data. There are a whole bunch of other electoral college calculations going on at other sites. Even sites with a clear McCain bias, like Real Clear Politics, shows nearly the same figures.
The one and only electoral college map I've seen that shows a McCain lead is the silly map of Bill O'Reily's. But by O'Reily's tortured logic, PA is a tie, even though Obama is polling +20 in some PA polls. VA is a tie, as are OR, MI, WI, etc. States that Obama is leading by comfortable margins. States that McCain has conceded and pulled out of. O'Reily's map is bias to the point of hallucination.
Yes, yes, yes. It's not over till next Tuesday, and anything could happen, and we all gotta make sure to vote, and make sure the Republicans don't steal enough votes to swing the election. But really, right now, today, McCain doesn't stand a chance.
Of course internal memos from the McCain campaign are going to say the race is tied (or they're ahead) when all the polling indicates otherwise. That is just to maintain morale amongst the staff and true believers and to keep them from just giving up. Would you be more or less likely to work hard for the next 6 days if your boss thought they were going to lose and you'd be out of work this time next week?
Since most of the election hasn't happened yet you have to look at the existing polls as outlining a couple of possible outcomes.
The trad likely voter polls have Obama barely ahead, so if it turns out that the youth vote, ethnic vote, and new voter vote does not turn out then the election likely ends up there.
At the opposite extreme you have Pew's unweighted poll of registered voters showing Obama up 18%. So if it turns out that Obama is successful in turning out an unprecedented percentage of registered voters, then his margin of victory would be somewhere in that range.
So volunteer (e.g. making Spanish calls into CO and NM has been a lot of fun) and vote and do what you can to advance scenario two.
@1 pointed out the only real "information" that any of this tells us. The Republicans are only interested in dumb shit jackasses who are easily manipulated. I'm shocked, I tell you, shocked!
His statistical methods are extremely good, that's why I read his stuff and what drew me to the site in the first place. He's very transparent about what he's doing with the numbers as well.
If I do have a fear, it is that Caribou Barbie brings out "new" voters; i.e., not formerly disenfranchised minorities, etc., but the 40% or so of ignorant rubes who typically *never* vote.
Could enough of them crawl out of their dens to influence the outcome? The implications are too horrible to consider. I still want to believe this is unlikely.
Too bad the slight uptick for the GOP reversed today and they dropped again.
Comments are closed on this post.