New Recall Poll seriously flawed

[jeter]

Okay for all you ready to jump off a bridge. Consider this, CNN/Gallup poll is notorious for running bad polls that favor Republicans (especially Bush supported).

The two most famous:

Bush and McCain tied in New Hampshire the day before the primary. McCain won the primary by 17%.

Bush ahead of Gore the day before the national election by 7%. Gore won the pop. vote by 1%.

In California, Gallup predicted that the election was within the margin. Bush ahead by 2% or Gore ahead by 2%. Gore won by 13%.

Why are their numbers like this? Simple. It's their methodology. They simply count more republicans than democrats.

Nationally, voter registration breaks down like this: Democrat 30 to 35% of population. Republican 25%. and Independant about 40 to 45%.


But in their polls they count it more like this: Republican 35 to 40%, Independant about 30 to 35%, and Democrat about 25 to 30%.

You see they undercount independants and Democrats. And overcount Republicans. That is why their polls favor the GOP.

I have written Gallup in the past and asked them about this. They responded to me by saying that this reflects the "probable" voter and not just the registered voter. Their argument being that the republican is more likely to vote than the democrat or independant.

Bullshit. There may be a kernal of truth. But not enough to justify these wild swings, that 9 times out of 10 benefit the GOP.

Okay, now with this most recent and famous recall poll.

Of California's "registered" voters, and this is a very interesting position as well. Since as we all know, the "registered" voters don't include many blacks, asians or hispanics. These groups make up more than 50% of California's population.

So already discluding these people, 55% of Registered Voters want to "remove Davis from office" whereas 41% do not.

According to the same poll of "probable voters" 63% want to remove him, whereas 35% want to keep him. 8% MORE want him gone, whereas 6% FEWER want him to stay. Using the same poll numbers.

So not only are they not counting the blacks, hispanics, and asians that are not registered. But they believe that enough democrats WILL NOT vote as to create 13% disparity between the poll of registered voters and "probable voters."

Look at the favorabilities:

Arnold has a favorable opinion among 63% of "probable voters;" 30% unfavorable.

McClintock has a favorable opinion among 62% of "probable voters;" 20% unfavorable.

Cruz Bustamante has a favorable opinion among 37% of "probable voters;" and 54% unfavorable.

There is no way that these numbers are true. With 50% minority population, two republicans who supported propositions against Affirmative Action and Hispanics have only 20% and 30% unfavorables. Whereas the Democrat has unfavorable of 54%? Not possible mathematically.

Finally the debate: Who won according to this poll?

Arnold got 23%; Bustamante got 12%; Huffington got 4%; McClintock got 43%; and Camejo got 10%.

Which means that 66% thought a conservative won. Whereas, 26% thought a left of center candidate won.

This again is impossible. Even in the debate Bush beat Gore (the second) 30 to 40% thought Gore won, whereas 50 to 55% thought Bush won. This is further proof that the numbers are skewed in favor of the Republicans.

There is no doubt that the GOP have gained some steam as of late. The nonesense about delaying the recall, the licenses and sam sex marriage mess. And the fact that Bustemante and Davis can never seem to be on the same page has hurt our momentum. But I think the numbers are close. Within the margin at very worst.

So stay the course. I believe we may still win. If Arnold's numbers were this good. Why has he gone on an attacking rampage? Why not be the likeable actor? His numbers aren't this good that's why.

I find it funny that these are the numbers though that get widely reported.

Check out the whole poll for yourself

www.usatoday.com/news/polls/tables/live/2003-09-28-recall-poll.htm

[MoonRiver]

1. CNN has gotten a little bit better in their reporting, but

it will never be a "trusted" news source until it uses a more credible pollster than Gallup. Gallup is a despicable rightwing shill pollster. If you sign up for a series of their online polls at THEIR invitation, as I did, and state you are a liberal Dem, as I did, expect to never hear from them again, as I didn't. :thumbsdown:

[realFedUp]

2. I think this was skewed to get McClintock out of the race

Republicans are putting the pressure on McClintock
to get out and making a poll seem like Arnold
is way ahead of him puts more pressure on him.

The race is undetermined...Arnold doesn't even
know who his voters might be and his campaign
can't do the kind of get out the vote that
the Dems or even McClintock can do on Oct. 7.

Don't believe everything that comes out of
CNN...you haven't before. :-)

[Hell Hath No Fury]

3. Thanks for those numbers and the info!

The "approval" numbers are the dead giveaway -- no way in HELL an ultraconsverative would poll that high unless there was an abundanec of conservative voters in that poll. And the winner of the debate numbeers -- hell, just call up 700+ "voters" in Orange County and ask them what you think, why don't you.

I knew something was stinky. :)

[knight_of_the_star]

4. They also aren't taking into account outside factors

A lot can happen in one week. Who knows? Arnie might take a headlong dive and the connections between him and Enron be made public. Anything could happen.

[0rganism]

6. What are the ties between Arnie and Enron?

If you can give me a one sentence summary, or point me to a link, I'd be interested in seeing that connection.

[Terwilliger]

8. I don't know specifically

Supposedly, Arnie met with Ken Lay about the state's energy concerns a couple of years ago. I know that Arianna Huffington brought it up, but I don't have any links.

[0rganism]

13. Found one at truthout... a May 24, 2001 session with LA rethug leaders

http://www.truthout.org/docs_03/081903G.shtml

Dunno if this is particularly damning stuff or not yet...

[sonias]

5. I'm still holding out hope

I want to see polls become completely unreliable. I still think with people ignoring phone calls (fearing telemarketers) and the fact that polls are not reaching a big voting segment (minority voters) you have unreliable numbers.

I want to believe that the Oct 7th CA race will put a stake in these phony polls.

Sonia

[SoCalDem]

7. We need to FLOOD the McClintock offices with SUPPORT

Tell them we know that the poll numbers are skewed to get him to quit.. We LOVE YOU TOM... Pleeeeeas don't let that android bully push you around :):):evilgrin:

oh.. an I would not advise and small aircraft travel.. forewarned & all :)

[PROGRESSIVE1]

22. "To the Honorable Governor Thomas McClintock"

This is how you should address you emails of support to Mr. McClintock!

[WCGreen]

9. In your normal run of the mill off year election,

I would tend to agree with the Gallup organization. Republican voters do tend to vote in almost all of the elections. Look at the voting turnouts in suburban areas vs. inner city neighborhoods and you will see why they jump to this conclusion.

I would, however, take the Gallup organization to task for the very same reasons you citied, they do not take into consideration any thing out of the ordinary.

For instance, in New Hampshire, I's wager a fair amount of money that they poled among republians only and did not factor in independents and democrats switching just to vote for McCain.

You can't even beging to consider what will happen in this election simply because it is in unchartered territory.

Polling registered voters would be the best way to do it since you can't be sure, based on prior models, who it is who is going to be voting.

I think it is much tighter than anyone could imagine and I would believe that AS has pulled closer to CB but, given the ethnic make-up of Calyfornia, I would imaging CB has to have an edge.

[PROGRESSIVE1]

19. My brother lives in SoCal right now.....

according to what he told me about the polls by the local stations
Bustamante is STILL in first place to replace Davis and the Yes/No question on the recall is statistically tied!

a local NBC affiliate poll:
Bustamante: 37%
Schwartzenegger: 29%
McClintock: 19%
Camejo: 4%

Recall:
yes: 50%
no: 46%

CBS affiliate:
Bustamante: 36%
Schwartzenegger: 27%
McClintock: 16%
Camejo: 5%
Huffington: 3%

They did not have a poll asking who would vote to recall Davis.

I would NOT trust the Gallup Poll!

:kick:

[jobycom]

10. Great analysis, thanks

Mostly this is a compliment and a kick for a great post.

[Robin Hood]

11. So by skewing polls, they are in fact vote tampering.

By trying to convince people that the majority of Americans would vote this certain way, and that must be the best way to vote, if you're a sheeple.

[West Coast Democrat]

12. I was surprised by the latest Gallup Bush Poll

...the one where is shows all of the Democratic candidates neck and neck with Bush. I smelled the tide turning against Bush; Gallup is usually the last poll to admit that the Bush administration is losing popularity.

[PROGRESSIVE1]

20. VERY TRUE!

Bush is in bad condition if the Gallup and Fox "News" polls show him tied with the Dem's!

[kentuck]

14. 18+ over in the poll...almost one/third not registered to vote..???

<snip>
Results are based on telephone interviews with 1,007 Californians, aged 18+, conducted September 25-27, 2003. For results based on this sample, one can say with 95% confidence that the margin of sampling error is ±3 percentage points.

For results based on the sample of 787 Californians who are registered to vote, the maximum margin of sampling error is ±4 percentage points.

Probable Voters include a subsample of 581 survey respondents deemed most likely to vote in the October 7th California recall election, based on their level of interest and intention to vote. This definition assumes a turnout of 51% in the recall election. For results based on the sample of probable voters, one can say with 95% confidence that the margin of sampling error is ±4 percentage points.


This poll was conducted in both English and Spanish -- 73 out of 1,007 interviews were conducted in Spanish.

In addition to sampling error, question wording and practical difficulties in conducting surveys can introduce error or bias into the findings of public opinion polls.

[jeter]

15. Thanks for the support guys

I think this is important. I know it's not proper online etiquette, but i'm giving myself a bump.

[dweller]

16. excellant analysis, and you've plugged the skewers

:thumbsup:

:kick:

dp

[CatWoman]

18. and I'm giving you another

:kick:

Thanks, Jeter :D

[spanone]

17. The Arnold polls make no sense whatsoever

I don't trust them. This could be a prelude to 2004.

[Hell Hath No Fury]

21. Keeping it bumped!

This is important info for CA voters!

[WaterDog]

23. The thing I worry about

is the BBV factor and how polls like this make it look feasible when they screw our votes.

[farmboxer]

24. It Appears As If Republicans Are Planning On Fixing The

elections with their Republican manufactured and programmed voting machines. The stack the polls in their favor will help cover up the fixed election. Sounds like Georgia and Florida again.

[David__77]

25. Thank you! You are right!

I was trying to get some traction pointing out a few of the poll's weaknesses, without much succcess. People were too busy arguing over defeatism and whatnot.

I would like to see the poll's questions regarding the recall itself. Asking "do you favor removing Davis from office," believe it or not, will get a different result than "will you vote to recall Davis from office." These subtle ambiguities are key in a race like this. I'm not a pollster, but if I were, I'd be studying this one closely after the fact to see what went wrong/right with polling.

We know, however, that it stands fairly close.

[VOX]

26. Excellent post, jeter. We need to keep this kicked...

so those who might be getting demoralized will buck up. This poll is utter BS, as sort of declaration of victory before the fact. It's an old and crummy strategy.

Thanks for posting the numbers --

:kick: