A question about 538's expected numbers over the next few days.

[FVZA_Colonel]

Is it the case that they are predicting McCain might have a 1 to 2 point lead in the next few days?

I've been having real problems getting my internet connection to work for any real length of time over the past hour, so I haven't had a chance to look on 538 myself (I'll be lucky if this posts.) I appreciate your help.

[dkf]

1. Zogby has McCain up by 4. Ugh.

http://www.zogby.com/news/ReadNews.dbm?ID=1548

[samuraiguppy]

2. zogby is a liar n/t

[dkf]

4. The polls are all going in that direction.

We only think the polls are right when we are winning.

[writes3000]

11. If a majority of polls show McCain ahead, it will be impossible to deny.

But to cherry pick one poll and then claim we're all in denial is at best premature. At worst...I'll call it suspect.

Have a good night.

[dkf]

13. I'm pretty sure the Gallup and Rasmussen tracking polls will be there soon.

Maybe not up by 4, but maybe 1 or 2.

[Josh]

3. While those numbers certainly aren't good I take Zogby with a truckload of salt

Especially when their polls lie like this:

"Zogby International was the most accurate pollster in every one of the last three presidential election cycles, and continues to perfect its telephone and interactive methodologies using its own live operator, in-house call center in Upstate New York, and its own secure servers for its online polling projects.

In the 2004 presidential election, not only was Zogby's telephone polling right on the money, its interactive polling also nailed the election as well."

This after they had Kerry with 312 electoral votes for 2004. So their data was way off from the official results and yet they still claim to have "nailed it." Where I come from (the Land of Reality), that's called "lying."

[dkf]

5. But why would they skew the poll for Kerry in '04 and McCain now?

Doesn't make sense.

If anything he's simply wrong.

[Josh]

23. Right, that's what I'm saying

I'm saying his polling sucks, not that he's biased. :)

The fact that he then *lies* about his previous poll results is enough to discount him all together.

[writes3000]

6. INTERACTIVE polls? The page you linked asks people to sign up to take their polls.

I'm sorry but this is WAY too easy to mess with their polls.

If I sign up and claim I'm a Republican but I'm now voting for Obama - it skews their poll.

Nope. Way too faulty and frankly, they should know better.

[dkf]

7. If anything an interactive poll probably shows enthusiasm.

And considering there are more Dems than Repubs, you'd think an interactive poll would skew in our favor.

[writes3000]

8. It's a completely un-scientific sample. What if some church group told everyone to sign up?

How would they know? It's not a random group at all.

And I bet if you look back at their interactive polls of late, those polls have a more pro-McCain bent than other polls taken at that time.

[dkf]

9. I used to participate in those polls.

But I do think that in the end they categorize us and then skew it to their view of the population.

[writes3000]

10. If it was truly the most accurate way to poll people, more pollsters would do it.

Just get 5,000 people to agree to be polled via email and voila!

I strongly suspect the reason why more pollsters don't do it is because it's not random at all.

Online surveys give you results like the AOL polls or the CNN polls. They are samples you cannot trust as much because the respondents come to you.

[dkf]

12. Polling has always been an inexact science.

But the actual number may not be as important as the trend from poll to poll. I think most here would agree that things are trending toward McCain.

[Teaser]

20. wrong

Polling is not an inexact science. Done properly, it is simply statistics, which is *very* exact.
Interpretation of polling is inexact.

And you repeatedly imply that an internet poll has *some* validity. Justify that to me, statistically.

[dkf]

21. Just seems to me that the polls we were hearing didn't bear

out in the primaries.

Sure, someone tended to get it right, but never one outfit consistently.

Zogby hit that Presidential election way back I know.

Its becoming a case of the stopped clock.

[regnaD kciN]

14. Zogby is an ONLINE poll. Ignore...

:eyes:

[Marsala]

16. Zogby's useless online poll has annoyed us all for months n/t

[Teaser]

19. It's an internet poll.

.

[regnaD kciN]

15. To answer the REAL question...

It's pretty simple. From past experience, each candidate gets a bump of around 4.5% just after their convention, which then subsides. What makes this year weird is that the conventions were so close together, so the decline in the Democratic convention bounce overlaps the Republican convention bounce, making for a steeper apparent shift, but the usual result is that, when all is said and done, the result coming out of the conventions is about the same as that going in.

Nate at 538.com has been following the tracking polls, and has a rough formula to determine what the differential each day out of the three-day tracking polls actually was. In the past week, the RNC had little effect through Wednesday, with Obama still holding the lead. On Thursday (taken after the Palin-mania of the night before), Obama still held a slight lead in both polls, but it was down significantly (from +7 to +1 in Gallup). On Friday (polling taken after McCain's acceptance speech), it appears that McCain won the day by 2 points in Gallup, but Obama continued in the lead by 1 in Rasmussen. (I will note that, despite all the rhetoric about Palin's effect on the race, it was actually the polling taken after McCain's speech, not hers, that went in the G.O.P.'s favor, suggesting to me that Sarahbou won't necessarily be the "game-changer" they so expected.)

Now, the results coming out tomorrow morning will be made up of the average of the results for Thursday, Friday, and whatever the polling was on Saturday. (TRANSLATION: It will be the first day where all the polling took place after Palin's speech, and therefore entirely in the time period of any RNC convention bounce.) We know that Gallup had Obama up one on Thursday and down two on Friday, so, if Saturday's polling gives us a tie or McCain lead, he'll be in the lead in the overall tracking poll as well. Rasmussen has had Obama winning both of the last two days, but by very narrow margins, so he might maintain a narrow lead, or a big polling day for McCain might put the latter in front. Remember, of course, that Obama has a tendency to poll weaker on weekends, when more young voters are going to be out and about, so the people who get contacted are more likely to be older and in McCain's demographic -- so it's possible that the Saturday polling will favor the Republcans more than would normally be the case, temporarily giving McCain a bigger lead.

According to studies I've read, a convention bounce generally reaches its peak 48 hours after the convention ends (in this case, Saturday evening), then will slowly starts to decline. Accordingly, it won't be too easy to tell how this has effected the race until at least after Saturday and Sunday results have passed through and fallen out of the average, which would be for the results released next Thursday. On the face of it, it would appear probable that McCain will grab a narrow (or even not-so-narrow) lead in both polls released this morning, possibly build on them the next day or so, then start to slowly drop back thereafter. My guess would be that, unless McCain has somehow managed to permanently change the dynamics of the race, a week from next Tuesday (9/16) will find Obama back in a narrow lead.

Hope this helps.

[FVZA_Colonel]

22. Thank you, I really appreciate this.

I was just able to get a steady internet connection again, so I'm going to go to 538 and check out the information myself.

[nerddem]

17. north dakota is +3 obama.

i'll let you extrapolate from there

[VolcanoJen]

18. I thought 538 was predicting a McCain +4 bounce?

Remember... a bounce is a bounce. Bouncing balls lose momentum, by definition.

[malik flavors]

24. I think they were predicting he'd have a 1 to 2 pt. lead, but it seems it might be greater than that

I'll be interested to see Nate's analysis of the latest polls later today.