New Gallup and ABC News Polls

[millwx]

ABC poll:

Among registered voters in this poll, 38 percent describe themselves as Democrats, 30 percent as Republicans and 28 percent as independents[/b]

A little Dem-heavy, but not severe (like LA Times).  Just further convinces me that calling it a tie is probably the most close to correct.  Of course, even at a tie that does indicate some shift back towards Kerry after Bush's "Reagan bounce".  So, I'd expect the next round of polls (those with good "balance", that is)... including Gallup, if its timeframe is late enough... to show Kerry +1 to 3%.

[The Vorlon]

I take it ABC does not weight by party ID?

No, unless any party polls 40% by party Id, they set a max.

[MODU]

Agreed Blue.  It does seem to be off somewhere.

[Smash255]

ABC is +8 Dems, but from the 2000 exit polls whichthe Dem adv was +5, its really not off all that much, and without knowing the TIPP breakout its hard to say.  For example if TIPP was Dems +2, its about the same off as the ABC poll, but in the other direction.  Regardless a difference of ebing +3 to heavy for the Dems isn't enough to have that much of an impact especially when the extra adv  has more to do with oversampling of Independents and Undersampling of Reps rather than Oversampling of Dems & Undersampling of Reps.

I stated in the other thread 2 possible reasons could tell the differemce in the polls

1.  Undecided voters,  Undecided voters tend to break rather heavily for the challenger in incumbent elections.  ABC pushes the undecidedes more than TIPP, and has a much lower ABC total than TIPP.  As a result of the smaller amount of undecideds and the undecidedes in Incumbent elections breaking for the challenger Kerry does better in the ABC poll than TIPP as the ABC poll keeps up with the trend of undecidededs breaking for the challenger

2.  Bush better week last Week, Kerry better in last coup[e days.  THE TIPP poll started a few days before the ABC poll & ended a fewdays prior to the ABC poll ending.  So if Bush was doing better right when the TIPP poll started before the ABC Poll started, but after the TIPP poll ended while the ABC poll was still going on Kerry was doing better it could explain the difference.  Also look at the Harris poll (partly taken last week) & look at the differences in the Rasmussen polls over the last week from Bush +3 last week  to Kerry +3 in the latest Rasmussen poll

[The Vorlon]

ABC is +8 Dems, but from the 2000 exit polls whichthe Dem adv was +5,

Actually +4

39/35/27 Dem/Rep/Ind (=101% due to rounding) in 2000 exit polls

But your point remains valid.

Theory #1 = unlikely, required ALL undecided/soft/lean to go Kerry.

2 to 1 for Kerry I'll buy, 100 to Zip I don't  

[Smash255]

ABC is +8 Dems, but from the 2000 exit polls whichthe Dem adv was +5,

Actually +4

39/35/27 Dem/Rep/Ind (=101% due to rounding) in 2000 exit polls

But your point remains valid.

Theory #1 = unlikely, required ALL undecided/soft/lean to go Kerry.

2 to 1 for Kerry I'll buy, 100 to Zip I don't  

Well I'm not saying all the undecidededs went to Kerry, bu if a majority of them went his way (lets say 70% or so) it could explain it, the difference then would be a much more understandable couple of points

[millwx]

Well I'm not saying all the undecidededs went to Kerry, bu if a majority of them went his way (lets say 70% or so) it could explain it, the difference then would be a much more understandable couple of points

Absolutely.  And let's not forget, we're talking about polls with MOEs here.  If they end up, after all reasonable analysis, still varying by a couple percent, big deal... there's nothing "wrong" with either of them.  That's why I give the ABC/WP poll some credence... it's sample, though perhaps a bit skewed, was not out in left field (like the LA times poll).  And for those who raise questions because of the terrorism question, keep in mind, it was posed to only half the sample (larger MOE) and, really, even it is not obscenely off... compared to the last poll that asked the same question of the same people (a Harris poll in mid-May) there was only a 4% shift (a change of 8%, but that's only 4% of the people changing their mind).

The ABC/WP poll STILL does look a bit Dem friendly in its sample, IMO.  But I, for one, see no glaring errors.  The Gallup Poll (due out today) MIGHT help clarify whether the ABC/WP poll is on target.

[The Vorlon]

A good validity check on the National polls is also the next round of state polls.

While there are certainly state to state minor ebbs and flows, in general the state polls move with national ones.

I Kerry has gained 6% nationally, the Michigan should roughly speakingmove from +4 Kerry to +10 Kerry for example.

Now the change might not be exactly 6%, but it will very likely be between say 4% and 8%.

Waiting For Gallup