Why is Gallup showing movement towards Obama while Hotline is tied?

[LiberadorHugo]

Wasn't Hotline the only poll that had O ahead at one point?

[WeDidIt]

1. That's a concerning concern

I'm now concerned.

[Bucky]

2. Maybe Gallup has bigger resources and produces more reliable data

[geek tragedy]

3. Hotline had a really odd sampling day enter

their data. Obama will bounce back.

[Drunken Irishman]

4. Hotline has a huge MOE.

[Schulzz]

5. If Obama leads Rasmussen by 6 the race is NOT tied. nt

[Sodan]

6. I look at it like this:

Hope Gallup is right, hope Hotline is wrong and go with Rasmussen and Ipsos/McClatchy (+6 & +7)

[mohc]

7. Why does Hotline have Obama tied among men...

...and then only leading overall by 1% when we know the gender gap would mean at least a 10% Obama lead if he were tied with men? There just seems to be something off with their sample recently.

[totodeinhere]

8. The latest Zogby tracking poll has it Obama 47%, McCain 45%

http://zogby.com/news/ReadNews.dbm?ID=1575

[FrenchieCat]

11. Zogby?

Notice Zogby's reliability ranking -

[FVZA_Colonel]

12. If even Zogby can't fuck with things enough to put him behind,

then I think we're doing ok for now.

[ClarkUSA]

13. Zogby's polls have always been screwy because of their weighting values.

Besides, national polls, good and bad, are meaningless. The only ones that count are battleground state polls from
reputable pollsters (which doesn't apply to them all). Furthermore, internal polls are the ones Team Obama is working
off of -- and the best indicator of what those says is to track which states the campaign is investing in and which
states Obama/Biden are visiting together and separately.

[thesubstanceofdreams]

9. Party weighting

Just two days ago hotline CHANGED their party weighting. I don't recall the exact numbers (nor could find them in their website) but the number of democrats is now republicans + 2%. This is absurd on its face because even in 2004 there was a 3% differential and everybody agrees this year the gap is going to become much larger, not smaller.

Frankly I just think there was a lot of pressure or money there as it doesn't make sense that they suddenly changed their methodology to one that doesn't stand the most basic analysis.

[ClarkUSA]

10. Because Hotline is owned by Republican partisans who used to be D.C. operatives?

And lest we forget, Chuck Todd was head guy of Hotline before he jumped to Universal NBC.

[kennetha]

14. I THINK it has to do with party ID

and whether the pollster weighs the sample or not. Some of the polls that have Obama way ahead, show Party ID of like 50 Dem, 39% Repugnant. (I think that was a Survey USA poll of Pennsylvania, if I recall). SOme pollsters use historical party ID to weight samples. Some let them stand -- on the assumption that party ID is shiftable and indicates enthusiasm, etc.

If you deploy these two different methodologies, you could get widely different results from the same raw data I think.

But I'm not expert and could have this entirely backwards.