Sarah Palin

[antlers]

I think if you look only at the population that was “undecided” before the conventions, Palin helped McCain a lot with both genders.

[Anaxagoras]

Unlike Lou Grant, America likes spunk.

They’ll vote for Jimmie Stewart over Truman every day.

This is an excellent summation of why I hate this country so very very much.

“Well, she’s an evil person, stupid, and ignorant, but I gotta say… She’s got spunk!”

You know who else had spunk?

[Bahimiron]

You know who else had spunk?

Jesus?

[Jazar]

Punkey Brewster?

[Inuvix]

This is an excellent summation of why I hate this country so very very much.

Once your hate reaches the “very very” stage, you should consider moving.

[divedivedive]

Punkey Brewster?

Pretty sure you meant ‘Spunky Brewster.’

A ha ha. Thank you folks, you’re really too kind.

[jeffd]

But predicting it ahead of time, folks didn’t seem to see quite so much of a McCain bounce.

Nate Silver at FiveThirtyEight.com predicted almost this exact bounce.

[Phil_Stein]

Nate Silver at FiveThirtyEight.com predicted almost this exact bounce.

If this is almost the exact bounce he predicted, why has the chance of a McCain victory soared on fivethirtyeight.com? (It’s over 50% now - I don’t have a history, but it was much lower a week or two back).

His site aims to predict who will win (or, more precisely, the odds of each party winning). Foreseeable events (including bumps from conventions) should be factored into his model, or his model isn’t very good.

FWIW, his models seem reasonable enough to me (though not necessarily perfect), from what I’ve read on his site. I think you’re probably mistaken when you state that he predicted this exact bounce (particularly the duration of the bounce).

[jeffd]

I don’t agree. Obama’s numbers didn’t jump much after the Democratic convention, and what bounce he did get was ephemeral. If the bounces both candidates received were typical, they’d cancel each other out and Obama would still be leading. However, the Palin pick is two weeks old, and McCain’s chances of victory continue to rise. For a change, I think the VP pick matters to a statistically significant portion of the electorate, and that’s why were seeing a more sustained increase in support for the Republican ticket.

No.

You don’t seem to have an understanding as to the origins of the “convention bounce.” It’s largely an artifact of press coverage focusing on a single candidate, usually positively.

First - Obama did enjoy a convention bounce. During the convention, and during the weekend afterward, his polling numbers went up. This bounce was blunted somewhat by the timing of the Sarah Palin announcement (a brilliant move on McCain’s part), and then by the GOP convention starting.

McCain’s bounce is slightly bigger than Obama’s because his convention went second. Nothing has happened in political terms to deflate his bounce, so it’s a bit larger and a bit longer lasting. We would absolutely not expect these things to magically cancel out, as you claim.

[jeffd]

Phil: Second graph in this post: http://www.fivethirtyeight.com/2008/09/patience-poll-watchers.html

Nate’s prediction is based on current polling. Current polling has McCain with a slight lead, so McCain’s odds of winning of the election if it were held today are slightly greater than Obama’s.

[Dejin]

If this is almost the exact bounce he predicted, why has the chance of a McCain victory soared on fivethirtyeight.com? (It’s over 50% now - I don’t have a history, but it was much lower a week or two back).

His site aims to predict who will win (or, more precisely, the odds of each party winning). Foreseeable events (including bumps from conventions) should be factored into his model, or his model isn’t very good.

IIRC he originally took his estimated convention bounce into account and tried to make victory predictions while smoothing out the bounce based on his model. However, people complained about his predictions not tracking the actual polls, so he removed his bounce estimate from his predictions. IIRC he had a post a while back where he specifically mentioned he was removing his bounce estimate from his overall estimate due to popular demand. So his estimates on winning percent do not include any leveling out for convention bounce.

FWIW, his models seem reasonable enough to me (though not necessarily perfect), from what I’ve read on his site. I think you’re probably mistaken when you state that he predicted this exact bounce (particularly the duration of the bounce).

His bounce estimate actually has a much longer tail. He’s not expecting McCain’s bounce to even out until the end of this month. According to his estimates McCain should be just below the height of his bounce now, with only a gradual drop for another week and a half, followed by a much steeper drop-off starting around 9/22.

Here is his actual graph along with a much more detailed explantion.

[Quaro]

His site aims to predict who will win (or, more precisely, the odds of each party winning). Foreseeable events (including bumps from conventions) should be factored into his model, or his model isn’t very good.

He thought so too, and initially there was a bounce correction, but the feedback was overwhelmingly negative so he took it out.

[Dave_Markell]

You don’t seem to have an understanding as to the origins of the “convention bounce.” It’s largely an artifact of press coverage focusing on a single candidate, usually positively.

Of course I understand that. Anyone with even a passing interest in the electoral process does.

First - Obama did enjoy a convention bounce. During the convention, and during the weekend afterward, his polling numbers went up. This bounce was blunted somewhat by the timing of the Sarah Palin announcement (a brilliant move on McCain’s part), and then by the GOP convention starting.

I never said Obama didn’t enjoy a convention bounce. What I did say was that it was small and ephemeral. To throw some actual numbers around, Obama’s bounce was a whopping 4%, less than half the modern historical average of 10%. I’d say Palin did more than blunt it somewhat–she chopped it by 60%.

McCain’s bounce is slightly bigger than Obama’s because his convention went second. Nothing has happened in political terms to deflate his bounce, so it’s a bit larger and a bit longer lasting. We would absolutely not expect these things to magically cancel out, as you claim.

McCain’s bounce was right around the traditional average. It wasn’t “bigger because he went second”, it was normally sized. What’s not normal this year is that Obama’s bounce was small. And that’s because of how the Palin announcement changed the dynamic of this race. If Obama had received the normal bounce, McCain’s equally normal bounce would have canceled it out and Obama would still lead. Instead, McCain is now ahead because Obama’s results were poor, despite one of the greatest campaign speeches ever given to an American audience.

[Dejin]

If Obama had received the normal bounce, McCain’s equally normal bounce would have canceled it out and Obama would still lead.

Not according to Nate Silver’s model. You can see how he developed it here, based on elections since 1952. According to that model McCain should still be up several points.

[Dave_Markell]

Not according to Nate Silver’s model. You can see how he developed it here, based on elections since 1952. According to that model McCain should still be up several points.

Yeah, that’s an interesting model. My numbers are from the AP, and are based on elections since 1964; his are based on those since 1952. This just goes to show that you can’t put blind faith in any single model, since they all have underlying assumptions and exclusions built in. It also doesn’t change the fact that Obama’s actual bounce was smaller than the level Silver projected (more than 6% predicted by Silver, 10% predicted by the AP; actual result 4%), which I again attribute to the Palin factor. Consequently, McCain’s gained more than just the 2% shown by the Silver convention bounce model.

[Dejin]

Yeah, that’s an interesting model. My numbers are from the AP, and are based on elections since 1964; his are based on those since 1952. This just goes to show that you can’t put blind faith in any single model, since they all have underlying assumptions and exclusions built in. It also doesn’t change the fact that Obama’s actual bounce was smaller than the level Silver projected (more than 6% predicted by Silver, 10% predicted by the AP; actual result 4%), which I again attribute to the Palin factor. Consequently, McCain’s gained more than just the 2% shown by the Silver convention bounce model.

I definitely agree Palin had a major effect over the short term and potentially over the long term. The Palin announcement occured on August 29, when Obama’s rise should have been in the middle and climbing, and it pretty much killed the bounce. Hard to say at this point whether that just means the Republican bounce started early or if there’s additional plus due to Palin. My sense is that Palin has meant at least a 1-2% real increase for McCain, but it’ll be difficult to say until the end of the month when a normal Republican convention bounce would have died off.

[JeffL]

He thought so too, and initially there was a bounce correction, but the feedback was overwhelmingly negative so he took it out.

I’m confused: he modifies his predictive model based on popularity of the model vs. what he believes is the most accurate model???

[Dave_Markell]

I definitely agree Palin had a major effect over the short term and potentially over the long term. The Palin announcement occured on August 29, when Obama’s rise should have been in the middle and climbing, and it pretty much killed the bounce. Hard to say at this point whether that just means the Republican bounce started early or if there’s additional plus due to Palin. My sense is that Palin has meant at least a 1-2% real increase for McCain, but it’ll be difficult to say until the end of the month when a normal Republican convention bounce would have died off.

I share your sense of a real McCain increase, and agree that we’ll have to wait and see if we’re right. Politics is a great spectator sport when you aren’t totally committed to either candidate. If you are, it’s either anxiety or joy producing (or both, as the rollercoaster zips on down the track).

[Dejin]

I’m confused: he modifies his predictive model based on popularity of the model vs. what he believes is the most accurate model???

Pretty much. If I were to take a guess, he probably got tired of answering e-mails asking him why the heck his model showed Obama doing so well when he was polling badly.

[Dejin]

I share your sense of a real McCain increase, and agree that we’ll have to wait and see if we’re right. Politics is a great spectator sport when you aren’t totally committed to either candidate. If you are, it’s either anxiety or joy producing (or both, as the rollercoaster zips on down the track).

Yeah election year is not good for my blood pressure. Also I waste far too much time tracking polls and reading commentary in spite of the fact that I have no real influence so it’s really just a waste of my time.