"What that "Awful" LA Times Poll Really Means"

[DemMother]

Good post from Donkey Rising (not from Ruy T.--I don't think he's back until tomorrow).

Two excerpts from long analysis:

"The Los Angeles Times poll released Thursday August 26th has created substantial consternation among democrats. Not only the mainstream media, but many pro-democratic writers and commentators have accepted the polls' apparent message that the sleazy attacks on Kerry's wartime record have been successful and have allowed Bush to overtake Kerry in the presidential race.

The bad news is that this perception has been widely accepted. The good news is that it's fundamentally wrong."

and...

"From the Bush campaign's point of view, the magnitude of the swift-boat fiasco becomes clear when it is recognized that a major goal of the August campaign was to put John Kerry on the defensive - to have him stumbling over his words, being pilloried in the press and firing his advisors. Instead (although the issue will now be muted by the theatrics of the Republican convention) it was Bush who was forced onto the defensive by the end of last week while Kerry weathered the attacks with an extraordinarily small decline in the level of his popular support.

Count on it, the Bushies are now very, very nervous. This wasn't the way they had it planned."


This second excerpt ties in to something I read earlier -- can't remember where -- that the Bush people were looking at their late convention as a triumphal celebration over the vanquished, demoralized Democratic opponent.

Also interesting is this from one of the comments on the site:

"Re: WSJ/NBC poll
The poll has Bush ahead 47%-45%, but this stat from the same poll, to my knowledge, has not: 45% of respondents said they voted for Bush in 2000, and 33% said they voted for Gore. 17% said they either did not vote, can't remember, etc. So Bush is in a statistical tie in a sample of voters amazingly skewed toward those who voted for him four years ago."

http://emergingdemocraticmajorityweblog.com/donkeyrising/index.php

[fujiyama]

1. Very interesting

that from a sample where 45% of the respondents voted for Bush in '00, and only 33% voted for Gore -- and Bush only wins by 2%!

[lancdem]

2. I like these grafs:

From the Bush campaign's point of view, the magnitude of the swift-boat fiasco becomes clear when it is recognized that a major goal of the August campaign was to put John Kerry on the defensive - to have him stumbling over his words, being pilloried in the press and firing his advisors. Instead (although the issue will now be muted by the theatrics of the Republican convention) it was Bush who was forced onto the defensive by the end of last week while Kerry weathered the attacks with an extraordinarily small decline in the level of his popular support.

Count on it, the Bushies are now very, very nervous. This wasn't the way they had it planned.

(BTW, people can't remember who they voted for in 2000?)

[jefferson_dem]

8. It's frightful to consider how many of these respondents

1. don't remember for whom they voted in 2000 or
2. claim they voted for chimpy when they didn't (bandwagon effect -- associating with the candidate who "won", thereby feeling like "winners" themselves)

[demgrrrll]

3. This is the second place I have read that the Republicans were hoping

that the Democratic candidate would be vanquished before their late convention. Hmmm.

[lancdem]

4. I think they were originally hoping he'd be vanquished

before HIS convention.

[BillZBubb]

6. If you were a Repug wouldn't you be hoping for that too?

That's hardly a revelation.

No doubt they thought Kerry's money limitations would work more to their advantage than it did. Fortunately for us, there was no good news for the Chimperor in August. The Swiftscum was all he had going his way.

[Bongo Prophet]

5. Polls can be used to demoralize opposition.

So don't let them get you down!
:dem:

[jefferson_dem]

7. Thanks for that upbeat and cogent analysis.

The point about the uneven sample is especially fascinating. And it may be even better news if one considers the general exercise of survey-watching. Perhaps there is a substantial "sample selection" bias in many of these polls in that more chimp-friendly people participate than chimp-unfriendly people. Natually, this would skew the results his way. I'm sure this is considered by the researchers and they as they weight the data if necessary and compute their findings. Still, something to ponder and be hopeful about.

[Stuckinthebush]

9. And there you have it friends....

The poll has Bush ahead 47%-45%, but this stat from the same poll, to my knowledge, has not: 45% of respondents said they voted for Bush in 2000, and 33% said they voted for Gore. 17% said they either did not vote, can't remember, etc. So Bush is in a statistical tie in a sample of voters amazingly skewed toward those who voted for him four years ago.


Always, always, always look at the methodology first.

[LisaL]

10. So, that means

Kerry actually would pick up 12 % of votes from people who either voted for Bush or didn't vote at all in 2000. So, actually, Kerry is gaining votes.
I think they are trying to make us panic with those polls with the way they conduct the polls and present the results.

[Stuckinthebush]

11. You got it, Lizzy

Sampling is everything.

[nolajazz]

12. There are books discussing only sampling methods for the entire book.

And sampling method is one of the most important topics in statistics, because it determines how reliable a statistical study, like an opinion poll is. That "45% of respondents said they voted for Bush in 2000, and 33% said they voted for Gore" really says something, because we know it is not even a good represention of the voters distribution in 2000. If we agree that nobody in 2000 voted for Gore will vote for Bush this year, the sample used in the NBC/WSJ poll was obviously skew
to Bush. I bet the "likely voters" in other poll (e.g. Gallup) did the same.

[eurolefty]

13. Some benchmarks

I don't know if this has been posted before, but here is a list of benchmarks for evaluating the Bush campaign.

-----------

Mark Mellman, Senior Strategist
Where Bush-Cheney Needs To Be
August 24, 2004

There are some basic benchmarks by which an incumbent's success can be measured as the campaign heads into the fall:

* The average winning incumbent has had a job approval rating of 60%. Indeed, every incumbent who has won reelection has had his job approval in the mid-50's or higher at this point. In recent polling, Bush's average approval rating has been 48%. President Bush must emerge from his convention having dramatically altered public perception of his performance in office.

* In recent years, when incumbents have gone on to victory, 52% of voters, on average, said the country was on the right track. Now, just 37% think things are moving in the right direction. Thus, President Bush must convince the electorate that the nation is in much better shape than voters now believe to be the case.

* Every incumbent who has gone on to be reelected has had a double-digit lead at this point.

* Following their conventions, the average elected incumbent has held a 16-point lead, while winning incumbents have led by an average of 27 points. Bush will need a very substantial bounce to reach the mark set by his successful predecessors.

* Incumbents have enjoyed an average bounce in the vote margin of 8 points. On average, incumbents' share of the two-party vote has declined by 4 points between their convention and Election Day.