Tracking Poll Mania: McCain Closes Gap, But Obama Still Leads

by: Chris Bowers

Fri Sep 05, 2008 at 13:59


I am going to need to enter tracking poll rehab after this election, but for now I am going to dive in head first and worry about the consequences later.

Today's tracking polls show Obama still ahead, but positive movement for McCain. Gallup moves from Obama 49%--42% McCain, to Obama 48%--44% McCain. Rasmussen moves from Obama 50%--45% McCain, to Obama 48%--46% McCain.

As those of you following the tracking polls closely already know, large movement was expected today, as Monday's enormously positive numbers for Obama were replaced by Thursday's post-Palin speech numbers. To show just how deep into the addiction I am, Haggai's daily tracking poll estimates for yesterday are Obama 47.7%--46.6% in Gallup (with a 1.5% MoE in both directions) and McCain 47.0%--Obama 45.8% in Rasmussen (with a 1.5% MoE in both directions). That averages to an exact tie.

So, it appears that Palin pulled McCain even, for one day, in the tracking polls. That is actually pretty good news, especially when one considers that Gallup and Rasmussen have consistently been 3.2%-3.4% more favorable to McCain than all other polls. Based on polling so far in the campaign, it is reasonable to assume that a dead heat in the tracking polls indicates an overall, narrow Obama lead. Even if his lead has shrunk, if Obama is still ahead after Palin, that is good news.

Now, the more sobering news is that tomorrow's tracking polls will be the first to reflect McCain's speech. While the speech didn't feel like a number mover, I still think it is reasonable to assume tonight's tracking poll surveys will be better for McCain than Tuesday night's, which they are replacing. A free hour of advertising has to be better than the giant goose egg that was Monday night of the Republican convention. Obama still held a solid, 4% lead in Tuesday night's surveys, but last night about 40 million people watched McCain's speech--more than watched Obama's. With yesterday roughly even, it is hard for me to imagine that Obama polls a 4% lead tonight, meaning that the tracking polls will continue to move toward McCain.

Returning to good news, we should take heart that it is unlikely polling will improve much from this point for McCain. The times immediately after conventions are almost always the high point for every single campaign. As such, the overall picture appears to be that McCain has closed the gap, but that Obama weathered the storm and still leads narrowly. I'll take it. At this point, I just want to win by any margin.

P.S. Hotline and CBS might also be instituting tracking polls. I'll look into that.

Update: Nate at 538 also tries to crack the individual daily numbers for the tracking polls, too. His estimate for yesterday is a bit more postiive than Haggai's, showing Obama up 1.5%. In the comments, Hagai sticks by his guns.  
Chris Bowers :: Tracking Poll Mania: McCain Closes Gap, But Obama Still Leads

Tags: , , , (All Tags)
Print Friendly View Send As Email
Tracking Poll Mania: McCain Closes Gap, But Obama Still Leads | 27 comments
come on (0.00 / 0)
let's see you dig in to the new tracking polls. Then we'll know you're committed. Or committable.
Don't tempt me (0.00 / 0)
Please! :)
[ Parent ]
Chris, I appreciate the hat-tip, but... (0.00 / 0)
...I just want to make clear that the "margin of error" in my daily number guesses is heavily dependent on all my previous guesses for each previous day, and not just dependent on the actual margin of error for the poll itself (not that you said otherwise, just clarifying for anyone who might think the two concepts had anything to do with each other).

In other words, the range values I provide for each candidate's numbers each day are "correct" if and only if (and boy, is that a BIG "if") my guesses for each of the previous two single days were correct.  I suppose I could fiddle with the ranges and provide a wider and more accurate "margin of error," but that would probably just defeat the purpose of the whole exercise.    

chris -- have you looked at nate's breakdown? (0.00 / 0)
he shows rasmussen's numbers from last night at Obama +2.  moreover, his model suggests that tuesday's numbers were quite good for McCain (+2), meaning that Obama has a bit of a "pad"  for tomorrow.  

i guess my only point is that, as everyone points out, it really isn't possible to be particularly accurate in breaking out a 3 day tracking poll, and that haggai's numbers may not be accurate.  

but beyond that, i'm happy to see that obama's numbers are remaining solid in the high 40s.  if he can maintain his level of support, even if mccain creeps up a bit, obama is in great shape.    

This one says Obama 51-46 (0.00 / 0)
But I suspect those Tuesday numbers are faulty. Unless of course this what Nate was talking about. Either way like you say its probably impossible to break it down accurately.
[ Parent ]
Ooops (0.00 / 0)
[ Parent ]
Nate is quite the statistical maven, but... (4.00 / 2)
...I think my numbers are likely more accurate than his.  I discussed the reason in this diary a few days ago (though that was before Nate posted any such numbers): rounding.  

If you look at Nate's numbers, you'll see some very, very large day-to-day swings.  For instance, his Rasmussen daily numbers show a 13 point swing from Obama to McCain on one day earlier this week, followed by an 8 point swing back to Obama the next day.  And his Gallup numbers show Obama expanding his daily lead by 12 points one day before plummeting back down by those same 12 points the next day, all in the middle of the Dem. convention...huh?

By guessing the daily numbers down to the first decimal place, I think my numbers provide a better guess at what's going on, simply by eliminating those wild day-to-day swings.

[ Parent ]
i can see what you are saying, though it sounded to me (0.00 / 0)
from reading nate's write-up that he may in fact attempt to take into account the issue of rounding.  though, i don't know enough about it to really have much of an opinion....
[ Parent ]
He might be taking some max/min ranges into account for multiple days (0.00 / 0)
That's what I took from his writeup (and he says he'll provide more details later, which I look forward to reading).  
[ Parent ]
I would tend to agree (0.00 / 0)
Yours do look more realistic. Then again the very reason they generally don't release a single days worth of numbers in any polling is because it is so volatile and open to wild swings so who knows the answer.
[ Parent ]
There's volatile, and then there's VOLATILE (0.00 / 0)
3 or 4 point swings in either direction from one day to the next is a certain level of volatility that can probably never be avoided, but 12-13 point swings is a whole other level.  
[ Parent ]
re: rounding (0.00 / 0)
I would be surprised if Nate is rounding to whole integers. In fact, I think at one point he notes that his Obama +2 estimate for today is actually Obama +1.7, which would suggest that he at least takes the tenths place into account. Is it the case, Haggai, that your numbers would reach an even closer approximation if you went to the hundredth, and so on? Of course, this may not be the case since we only have whole integers from Gallup/Rasmussen to start with.
[ Parent ]
Yeah, I'm pretty sure he's not just eyeballing it (0.00 / 0)
Which is pretty much what I'm doing.  Still, I'll be interested to see more details from him about his estimates.

Good question about going down to the hundreths, but I believe the answer is no, to the question of whether that would really give a better approximation.  I guess it might give a more precise approximation for the range of each candidate's numbers on any given day, but that could very easily be at the sacrifice of accuracy in getting a range that makes sense.  

Overall, I think my method--including my just recently decided idea of assigning the middle value of each range to each candidate as that day's number, as opposed to my previously more haphazard assignment of any number within the range, just depending on what "felt" right--seems to strike a decent balance between trying too hard to get the "right" daily numbers while also avoiding gigantic swings from one day to the next.

[ Parent ]
Have you discussed with Nate? (0.00 / 0)
Haggai,

Do you ever post at 538 and try to trade ideas with Nate about your different methods? It might be interesting and enlightening to us all.

[ Parent ]
No, I haven't posted over there (0.00 / 0)
But I'll keep an eye out for his explanation of his algorithm.  He says he just needs a bit more time to work with before posting the details.  Once he does, I'll look closely at his method and post something here.      
[ Parent ]
"Hotline and CBS might also be instituting tracking polls." (0.00 / 0)
We might have to distinguish between what they might be calling a "tracking poll" and the rolling average that constitutes the Rasmussen and Gallup numbers. I suspect that is the case for Diageo/Hotline while CBS seems more sporadic.
Has any.... (0.00 / 0)
Candidate in recent history outpolled the peak of their convention bounce on election day?  

I ask because, if McCain doesn't have the race at a dead heat or better by Monday (my guess is he'll still be a point behind, at least in Gallup, as the post-speech press coverage doesn't look too glowing), is there any historical precedent to not stick a fork in him?  

Good question (0.00 / 0)
In that scenario he would probably need some sort of external event or for Obama to crash and burn somehow.
[ Parent ]
Nixon 1972, Reagan 1980 & Bush 1988 (0.00 / 0)

[ Parent ]
Diageo has posted tracking numbers (0.00 / 0)
and it's Obama +6%, which too me seems too good to be true for a tracking poll post Palin's speech.

Are you guys going to include that in your composites?

Like I said, don't tempt me! (0.00 / 0)
LOL...but yeah, I imagine I will.  I can't access the Diageo numbers right now, but I think it should be pretty easy to start throwing something together for them within the spreadsheet I'm using.
[ Parent ]
Great :-) (0.00 / 0)

[ Parent ]
It still goes in the composite (0.00 / 0)
But the other two have bigger samples so are probsbly more accurate. Having said that, as Chris keeps pointing out, Gallup and Ras have tended to show smaller Obama leads than almost all other polling so maybe not.
[ Parent ]
I'll be happy if it's tied (4.00 / 2)
If, when the convention is all said and done, the race is about even, I think we should be relieved. It is normal for a candidate to surge after the convention and their post-convention bounce typically is their high-water mark. I think the GOP really has to talk the electorate out of voting Democratic this year, and if the numbers don't change dramatically after the convention it will indicate that they haven't yet done that, and they are running out of time.

By this time in 2004, Zell Miller had already savaged Kerry in prime-time (Palin's handful of sarcastic remarks are nothing compared to Zell's full-on assault), and we were knee-deep in Swift Boat Bullshit. This all came after Bush post-convention bounce. There is no indication that the McCain campaign is being run as well as the Bush campaign was. Bush was running on much more favorable terrain. And he still barely squeaked out a victory. So guarded optimism is the order of the day.

McCain's speech bounce (0.00 / 0)
I posted a comment on the Part 3 diary about McCain's speech that I thought the best part of his speech was the end.  But prior to that, he'd already turned me off on his policy discussions.  Same old GOP.  Forced childbirth, tax cuts to rich, etc etc.  Happily, McCain got that drivel in early enough in the speech for people to say, "Same old..." and turn him off/tune out.

Maybe the bounce won't be so big.

I have to admit that at the end I wanted to break into my old college fight song....

"Fight. Fight. Fight with all your might, Western win this game!"

Extra credit for the first response that correctly picks my alma mater. ;)

I'm amazed that more people saw McCain's... (0.00 / 0)
...speech than Obama's... must have been rollover from football... Does anyone have a breakdown by network?

From all accounts, his speech was gawd awful... I refused to watch it... too depressed over the week's events... but, it was on the TV, and when I walked past, I noticed the lime green background... I couldn't believe they'd be so stupid as to do that again (yes, I know, it was grass, but, they still blew it!)

I would have to imagine that his speech would have been a major letdown compared to Palin's...  He might actually lose some support post Palin...

REID: Voting against us was never part of our arrangement!
SPECTER: I am altering the deal! Pray I don't alter it any further!
REID: This deal keeps getting worse all the time!

From all accounts with me (0.00 / 0)
people at home loved the speech, some thought it was better than Obama's. I was kinda shocked by the response it got.  
[ Parent ]
Tracking Poll Mania: McCain Closes Gap, But Obama Still Leads | 27 comments
Donate to Open Left









QUICK HITS

Friends of the Earth thanks the OpenLeft community for the ideas you generate and your contributions to the progressive movement.

blog advertising is good for you
blog advertising is good for you
LIBERAL INSTITUTIONS
STATE BLOGS
SEARCH

   

Advanced Search

Home | RSS/XML Feed | Register | Contact

  © OpenLeft 2007-2009, Powered by SoapBlox