Eerily predicted by Tina Fey on 10/29/2005 SNL:

"The latest Gallup poll found that 66 percent of Americans think President Bush is doing a poor job in Iraq. The remaining 34 percent believe that Adam and Eve rode to church on dinosaurs."

Of couse when the poll is weighted in an extremely lopsided fashion, the results are completely meaningless.

Check out Real Clear Politics. It provides
some detail as to the poll participation:

- - - - -

Now for the numbers behind the numbers.
Take a look at the composition of the
respondents:

Total Respondents (Unweighted) = 936
Republicans = 259 (27.67%)
Democrats = 326 (34.83%)
Independents = 351 (37.5%)

Now look at the weighted sample:

Republicans: 223 (23.80%)
Democrats: 326 (34.79%)
Independents: 388 (41.4%)

- - - - -

It's seems that voters must be deserting
the Republican party in droves since the
2004 election. Or perhaps the sample was
skewed.

It would be interesting to see results
based on 1) likely voters 2) since Alito
was nominated

Personally, I think the results would be
much the same. But the right wing talk
shows are going to be leaning heavily
on the proper statistical sample - or
lack of it - in the next few days.

Geopolitical expert Tina Fay sure is funny. Must be why the show is doing so great with ratings.

2000 dead and what do we have to show for it, two lousy elections, a constitution, a working goverment, backbones of an army and eroding support for the terrorists? Isn't 2000 dead like, eh, worse than Vietnam?

Car bombs = US is loosing the war. Tina Fay says so, Howard Dean sais so, John Stuart says so. You have to be a dumt creationist not go understand that.

Gallup poll carried out Oct 28-30 (Libby was indicted on the 28th) has Bush at 41% - 2 percentage points HIGHER than prior to the indictment and during the Miers fiasco.

Alito was nominated on the 21st, hence not taken into account in this poll. Certainly the base will have been rallied more than under the Miers nomination. Expect the numbers to increase based on that in the next poll.

Alito was nominated on the 31st, not the 21st.

Dan clearly cribbed his headline from Wonkette...

Last Friday, the I Ching (ancient Chinese book of divination) predicted Bush's future political rebound, if he does the right thing. Talk about strange political bedfellows! Bill Kristol and the I Ching! That's why I love politics:

http://uselesstree.typepad.com/useless_tree/2005/10/friday_i_ching__3.html

Look, we've just gone through a bout with a couple hurricanes during which the press dove headfirst into fantasies about what was happening in NOLA... all of which they blamed on Bush, and none of which turns out to be true. At every opportunity in the last 6 byears and more, the press has been the left's house organ, and on every occcasion, the anti-Bush crap the press has been tossing, has proven to be flat out lies... pure fabrications.

So now they tell us Bush isn't polling well, with an obviously lopsided polling structure, and you actually take it at face value?

I must say I'm disappointed, Dan.

Why do Democrats participate in so many polls? Oh, they don't have jobs. Forgot. Sorry.

Can anyone with some knowledge of the subject explain to me why they believe that this re-weighting was wrong? (This is not a rhetorical question, I'm genuinely asking.)

Here's my vague, non-expert sense of the matter:

1. It is OK as a matter of sampling practice to re-weight, for example, in order to reflect the fact that your poll happened to get disproportionately few responses from one group and disproportionately many from another group.

2. CBS re-weighted the sample in favor of Democrats and against Republicans.

3. The objection raised here is not that any re-weighting in favor of Democrats would be bad, but that there doesn't seem to be a rhyme or reason to the numbers, particularly given that the sample already had substantially more Democrats than Republicans.

I am hung up on #3 here -- I grasp the objection, but would like to know a little bit more of the technical background there.

You all can do the math for yourself and see how meaningless this poll is.
I have always understood that the US population is about 40% Repub, 40% Dem, and 20% Ind (these are my own assumptions-you can substitute your own understanding below to calculate a more accurate total, if you wish).
According to the CBS poll, 77% of Repubs like the job Bush is doing, 30% of Independents do, and 11% of Democrats do (see the link to the poll).

The formula, then, is
.40 x .77+ .40 x .11 + .20 x .31 = .414, or 41.4% approval for the president.
If you don't like my estimates for Repub/Dem/Ind, substitute your own. We can all calculate the 'bogusness' of the poll for ourselves!

Steve

I can criticise President Bush for having any troops in Iraq, or for not having more troops in Iraq. I can criticise President Bush for having cut taxes, or for not yet having done away with the Death Tax. I can criticize President Bush for not spending enough on the environment, humanities, etc., or for spending anything on them.

What then does "disapproval of the President's policies" mean?

The best polling agency is

http://www.rasmussenreports.com/Bush_Job_Approval.htm

They do daily polls, and were the best agency when it came to predicting the 2004 election (split first place from over a dussin).

They show that Bush seems to have goten a slight increase after Alito, but nothing huge, although of cource the differences are too small to tell now.

CBS is a JOKE. The last election 37% self identified as Republicans, 37% democrats. How the hell do they get 24% Republicans? Fantasy?

Bush is in trouble yes, because of Iraq I think and maybe gasoline.

Hawing said that, the Republicans are still pretty strong. If you want to see naked evidence of liberal media bias look at how they are predicting the 2006 elections, compared to electronic markets:

about 77% probabillity Republican majorety both senate and house

http://www.tradesports.com/partners.jsp?ZID=null&AID=1&CID=3&page=trade&selConID=291849

If the press is objective, why are these guys not making a furtune betting against the republicans?

For people looking at polls try Mystery Pollster.com. The guy is pretty consistent in his work explaining how polls work et al. As of last night he had not gotten to Bush's poll numbers