What are McCain's chances of victory?

[Lunar]

What I can't wait for is when Nate Silver has a "Zogby" or "Rasmussen" moment.  Then we'll see how much he's really worth.

What is a "Rasmussen" moment?

The fact that Nate weights all of his polls based on their predictability ratings in the 2008 primaries is crazy. 

[Sam Spade]

What I can't wait for is when Nate Silver has a "Zogby" or "Rasmussen" moment.  Then we'll see how much he's really worth.

What is a "Rasmussen" moment?

The fact that Nate weights all of his polls based on their predictability ratings in the 2008 primaries is crazy. 

To be honest - a severe f-up (see 2000).

[Lunar]

What I can't wait for is when Nate Silver has a "Zogby" or "Rasmussen" moment.  Then we'll see how much he's really worth.

What is a "Rasmussen" moment?

The fact that Nate weights all of his polls based on their predictability ratings in the 2008 primaries is crazy. 

To be honest - a severe f-up (see 2000).

It's going to happen.  I'm starting to become more and more annoyed with how he does things.  Assigning undecided voters at the same rate they broke in the primaries?  Assigning pollster's weighting based on their error during the primaries?

[Mr.Jones]

@ Lunar, Selzer and Co is +0.75 while Columbus Dispatch is +8.00 . Which number predicts more accuracy, the higher number or the lower number ? Sorry I am a bit confused with the chart.

[Lunar]

@ Lunar, Selzer and Co is +0.75 while Columbus Dispatch is +8.00 . Which number predicts more accuracy, the higher number or the lower number ? Sorry I am a bit confused with the chart.

Pollster-induced error is a bad thing   Less is better.

[Lunar]

@ Lunar, Selzer and Co is +0.75 while Columbus Dispatch is +8.00 . Which number predicts more accuracy, the higher number or the lower number ? Sorry I am a bit confused with the chart.

Pollster-induced error is a bad thing   Less is better.

I should add that this list of error during the primaries is one of my biggest complaints about the 538 model.  Some of the best pollsters had trouble with caucuses and the low-turnout primaries, but that doesn't mean that they are bad general election pollsters.

[J. J.]

You still are 100% wrong if you're going to interpret that poll to mean that there is a Bradley-Effect of 3-5%.  You can't spin your way out of this one, it's just not true.  And we don't know where the truth lies (are people lying to the black interviewers or the white ones?).

  A difference between race of the interviewer does not mean that people are lying to Gallup and Rasmussen, unless you're going to say that Gallup is using all black interviewers!!  We assume that most interviewers used in polling are white, so if this DOES represent the Bradley Effect than your own logic indicates that there is no Bradley Effect in current polling.

Lunar, please don't words into my mouth.  I said, now repeatedly, that I thought it would be 1-2 points.  There was a pollster that noted the difference of 3-5 points in the responses.  The actual numbers from 2006 looked like roughly a 2-4 point spread (or including the really outer edges 0-7 points).

Do I think Obama will not perform as well as the polling shows.  Yes.  Enought to make a difference in a 7 point race?  Probably not.  In a 1 to point race, yes.  In a 2-4 point race?  Maybe.

I'm not going to argue with you about the existence of the Bradley Effect.  Just don't cite this poll as indicating that there could be a 3-5% effect.  I know very well what you believe and how powerful this effect is, but you're misleading everyone else when you talk about the existence of this poll which doesn't exist.

Lunar, only you could take a  1-2 point range and claim it's a 3-5 point range.

[Lunar]

I never did.  I never said you thought that and only you could keep persisting saying that I said this as a way to completely ignore my point. 

What you DID do was cite a poll claiming it said something it didn't.

[Sam Spade]

What I can't wait for is when Nate Silver has a "Zogby" or "Rasmussen" moment.  Then we'll see how much he's really worth.

What is a "Rasmussen" moment?

The fact that Nate weights all of his polls based on their predictability ratings in the 2008 primaries is crazy. 

To be honest - a severe f-up (see 2000).

It's going to happen.  I'm starting to become more and more annoyed with how he does things.  Assigning undecided voters at the same rate they broke in the primaries?  Assigning pollster's weighting based on their error during the primaries?

The two things you point out are big problems, I do agree.  After all, primaries and general elections are really apples and oranges.  If you want to weight pollsters for general elections, look at how they've done the past five GE or so, not based on one primary.

And if there's one election where I suspect undecideds may break completely one way or another, it's this one.

Finally, if there's a shift in the last few days or so of an election, his statistical model will invariably miss it.  That's where good pundit *gut instinct* is more important than statistics.

My pundit *gut instinct* has said for the past two weeks (since the bailout vote failure) that the race is essentially Obama +5 and would be somewhere in-between Obama +5 to +7 if the election were held today (remember what I said about the undecideds).

However, and I expect some disagreement here, the Ayers and Acorn stuff is starting to make an impact.  I suspect that as I sit here right now, the lead is around Obama +4, rather than the +5 it was last Friday.  

Obama has one main place to dampen any growth on that this week and that's the debate.  Also, if we see more economic turmoil, the numbers will move back to where they were.  Otherwise, if the GOP keeps hammering, expect to see continual tightening this week.  I expect Obama's people to fire back as well fairly soon, as they know quite well what I'm seeing - but we'll see what effect that has.

[Rococo4]

What I can't wait for is when Nate Silver has a "Zogby" or "Rasmussen" moment.  Then we'll see how much he's really worth.

What is a "Rasmussen" moment?

The fact that Nate weights all of his polls based on their predictability ratings in the 2008 primaries is crazy. 

To be honest - a severe f-up (see 2000).

It's going to happen.  I'm starting to become more and more annoyed with how he does things.  Assigning undecided voters at the same rate they broke in the primaries?  Assigning pollster's weighting based on their error during the primaries?

The two things you point out are big problems, I do agree.  After all, primaries and general elections are really apples and oranges.  If you want to weight pollsters for general elections, look at how they've done the past five GE or so, not based on one primary.

And if there's one election where I suspect undecideds may break completely one way or another, it's this one.

Finally, if there's a shift in the last few days or so of an election, his statistical model will invariably miss it.  That's where good pundit *gut instinct* is more important than statistics.

My pundit *gut instinct* has said for the past two weeks (since the bailout vote failure) that the race is essentially Obama +5 and would be somewhere in-between Obama +5 to +7 if the election were held today (remember what I said about the undecideds).

However, and I expect some disagreement here, the Ayers and Acorn stuff is starting to make an impact.  I suspect that as I sit here right now, the lead is around Obama +4, rather than the +5 it was last Friday.  

Obama has one main place to dampen any growth on that this week and that's the debate.  Also, if we see more economic turmoil, the numbers will move back to where they were.  Otherwise, if the GOP keeps hammering, expect to see continual tightening this week.  I expect Obama's people to fire back as well fairly soon, as they know quite well what I'm seeing - but we'll see what effect that has.

yeah I basically agree with that...especialy the undecided stuff.  there is going to be a major break one way or the other, unlike 2004.  I as a partisan dont pretend to know which way it will be.  But I think it will happen.

[TheGlobalizer]

You're taking the Bradley effect into too much prospect, JJ.  There was hardly any Bradley effect in states in the primaries, and in fact, he underpolled in some states.  Why do you think people who voted for him in the primary won't vote for him in the general because they realize he's black?  Seems a bit silly to me.

I'd submit that the Bradley effect is probably less pronounced for bona fide Democrats than for moderates or conservatives, and especially in socially conservative / racially divided states, such as those in play in this election.  I could certainly see a stronger general election effect than in the Dem primary.

But I think the Bradley effect is generally in its waning hours at this point.

Anecdotally, my girlfriend's mother lives in NC and supports Obama.  She wore an Obama shirt while walking around town (Greensboro) recently and got a lot of positive reaction to it.  Times are changing, and regardless of my soft support for McCain, I'm glad they are.

[Queen Mum Inks.LWC]

30-32%

[Swing low, sweet chariot. Comin' for to carry me home.]

with the stock market collapse halted, I voted 10-20%, though I would put them somewhere between 10 and 15%