Obama Opens Up a 9-Point Lead

[Naked Gents Rut]

and we know that post-debate, Obama got a decent bump (the margin should probably be closer to 6 or 7).

1. "We" don't know that Obama got a bump from the debate yet.

2. Obama's lead is only around 4% according to the latest polling.

Jesus, stop being such a predictable hack.

[Ramo]

Stop being such a predictable douche. There are 4 tracking polls out, and the average shows a small debate bump.

[Ramo]

Pre-debate Gallup had Obama +3; yesterday he was +8.

Pre-debate Rasmussen had Obama +5; yesterday he was +6.

Pre-debate Research 2k had Obama +5; yesterday it was +7, and today, it's +9.

Pre-debate Hotline had Obama +7; yesterday it was +5.

Note that the sample sizes for Research 2k and Hotline are about a third of Gallup and Rasmussen.

The one non-tracking poll that I've seen is the NYT poll, which saw Obama go from +3 to +5.

BTW, I don't trust GWU (they track over long time periods, and intermittently skip days), and Zogby Interactive is complete crap (an internet poll).

Average this, again, and you've got a couple points bump. Given the trend, it looks pretty likely that the average for today's polls will be in the +6 to +7 range, if not more. Both sites you linked to average over longer time periods to get more robust data (the tracking polls themselves do), but obscure short trends. Which I think is accurate as for as predicting the election, since bumps often dissipate. But I'm talking about the current state of the game.

[Naked Gents Rut]

Both sites you linked to average over longer time periods to get more robust data

No kidding.

[Ramo]

What the hell's your point then? The fact that I was talking about a debate bump indicated that I was referring to a much more local average in time. Hell, I even posted the robust average coming from one of the pollster.com guys in the graph I attached.

Your comments are just dumb...

[Naked Gents Rut]

You're the one crowing about a "debate bump" that isn't supported by any of the polls yet. Two points...

1. The tracking polls may be showing an upward trend for Obama, but that's no proof of a debate bump. The trackers won't have shed their pre-debate data until today. It's also going to be extremely hard to assign any pro-Obama shift in the polls to the outcome of the debate given that Obama has been trending steadily upward since the financial crisis kicked into high gear. The movement in the trackers you're harping on to prove Obama got a debate bounce could just as well be the result of statistical noise or fears over the economy rather than Obama's performance in the debate.

That's assuming there even has been a trend toward Obama in the past few days. The tracking poll results contradict each other on that issue.

2. Trying to determine how big Obama's lead is via tracking polls is ill-advised, given their small sample sizes and general volatility. There's a reason people turn to aggregate polls for an accurate estimation of the candidates' position in the race. Tracking polls are good for detecting trends but little else.

[Ramo]

1. All I said was that post-debate, he got a couple point bump. I never said that his bump is derived entirely due to the debate; you're reading that straw man into my post. Obviously other factors are involved.

2. Trackers are normal polls. They average over a few days to accommodate for small individual day samples. That's why I posted the 3 or 5 day averages (depending on the poll). Gallup and Rasmussen, in particular, have large sample sizes. I am using the aggregate of polls, and specifically referred to the trackers (plus the NYT poll) because I'm talking about a more local time period.

Seriously dude, this is a really dumb argument that you're trying to make.

[Ramo]

That's assuming there even has been a trend toward Obama in the past few days. The tracking poll results contradict each other on that issue.

You realize the absurdity of your post when you're talking about aggregates in one sentence, and contradicting polls in the next, right?

[Naked Gents Rut]

The aggregate of polls is still showing Obama at +4, not +6 or +7. Seriously dude, that was a dumb argument to make.

As a side note, both the Gallup and Rasmussen tracking polls show no change today. So much for the debate bounce.

You realize the absurdity of your post when you're talking about aggregates in one sentence, and contradicting polls in the next, right?

There's a big difference in the statistical validity of the average of five polls and the average of every poll. How do you identify outliers amongst five polls when one shows a loss, two show little change and two show a big gain?

[Ramo]

The aggregate of polls

An aggregate has a meaning only over a specific time period. I specifically didn't choose the time period associated with O+4, and I mentioned that. I repeat, you're making a really dumb argument.

As a side note, both the Gallup and Rasmussen tracking polls show no change today. So much for the debate bounce

Compared to Friday's release, there's a large bounce in Gallup. O+3 to O+8.

[Imran Siddiqui]

Yeah... the House Republicans were really just delaying things until McCain could get credit for the bill:

http://www.nytimes.com/2008/09/30/business/30bailout.html?hp

WASHINGTON — The House of Representatives was locked in a historic vote on Monday on a $700 billion rescue of the financial industry, in what would turn out to be the biggest federal intervention into private markets in the nation’s history.

With enough “no” votes recorded to defeat the measure, barring a change in position by some lawmakers, the markets were plunging early in the afternoon. Meanwhile, House leaders were trying frantically to turn some “no” votes into “yes” votes by pointing to the damage being done on Wall Street.

Just admit you were wrong now, Ramo.

[Ramo]

There's a big difference in the statistical validity of the average of five polls and the average of every poll.

What does that matter? I specifically used a narrow time period. That interval has a larger margin of error. What's your point?

How do you identify outliers amongst five polls when one shows a loss, two show little change and two show a big gain?

The two sites you referenced don't throw out "outliers" (unless it's from a disreputable firm). They normalize a poll by sample size, and treat them equally.

[Naked Gents Rut]

Just admit you were wrong now, Ramo.

He's incapable of it.

The two sites you referenced don't throw out "outliers" (unless it's from a disreputable firm). They normalize a poll by sample size, and treat them equally.

I never said they did. My point is that you can hope to reasonably estimate the true population mean when you have a large sample size to work with. 5 polls, on the other hand, is far too small a sample to draw any meaningful inference about the true mean.

That interval has a larger margin of error. What's your point?

My point is that your "6-7" point Obama lead is fiction.

[Ramo]

Yeah... the House Republicans were really just delaying things until McCain could get credit for the bill

I didn't say that. I said there were multiple concerns involved. Repeatedly. Hell, things substantially changed for Dems too (95 voted against the bill).

I'm not going to say that I know exactly what happened, but I doubt that McCain's role was terribly productive. That was my point.

[Ramo]

I never said they did. My point is that you can hope to reasonably estimate the true population mean when you have a large sample size to work with. 5 polls, on the other hand, is far too small a sample to draw any meaningful inference about the true mean.

What the hell are you talking about? We can draw an inference quite easily. The margin of error is larger, but the idea that the information is meaningless is bull****.