Articles and Analysis


Gallup: Confirms Rise in Bush Job Rating

The latest USAToday/Gallup poll out this morning (USAToday story, results, Gallup summary) confirms the modest increase in the Bush job rating in evidence in some (but not all) recent polls conducted in September.  Among all adults, Bush's job rating approval rating has increased to 44%, which Gallup characterizes as "an improvement compared with the public's assessment of his performance in recent months" and "his highest rating so far this year." 

The new survey also provides results to the generic Congressional ballot question among likely voters.  While the Democrats lead among all registered voters (51% to 42%), the result is dead even (48%44% each) among those classified by Gallup as likely voters.

This is one poll, of course, subject to the same random variation as any poll. The Gallup summary notes how "fluid" the Bush job rating has been ("measuring as high as 42% in mid-August, but dropping back to 39% earlier this month").  As has been Gallup's recent practice, the summary also includes a three-poll rolling average and a "smoothed estimate" based on the Samplemiser program developed by Yale Professors Donald Green and Alan Gerber.  Both estimates suggest that Bush's current job rating has increased but to a slightly lower level (42%) than the result of this most recent poll.

Clearly anticipating the first question many will ask, however, the Gallup summary also includes this important point about party identification:

The improvement in Bush's ratings appears to result from a more positive evaluation of him from all party groups, rather than a short-term shift in more basic party loyalties. In the current poll, 34% of Americans identified as Democrats and 31% as Republicans. In the prior September poll, when Bush had a 39% approval rating, 35% identified as Democrats and 30% as Republicans. 

Both the new Gallup summary and a companion analysis that suggest the "Republican strategy on terror . . . may be having an effect," are free to non-subscribers for today only. 


Alan Abramowitz:

The gap between registered and likely voters appears too large. Gallup's likely voter model has proven highly unreliable when used this far in advance of an election. Most notably, during the 2000 campaign, Gallup's likely voter results showed wild gyrations in support for Bush and Gore in the span of a few days.



Sorry, I don't buy it. I don't know a single person who approves of Bush.


You are so right. Gallup is paving the way to another stolen election, just like in 2000/2004. What has Bush done to make his ratings jump in the last month?

Advocate overhaul of the Geneva Convention? Turn Republican senate leaders and Colin Powell against him? Link Iraq to 9/11 in that "npn-political" speech? Plan for another October surprise in Iran or elsewhere? Sit by while civil war accelerates in Iraq?

Does anyone truly believe that those who disapproved of Bush in August now approve because the oil companies manipulate gas prices just before the election?

When will they stop drinking the kool-aid and see through the Gallup charade which comes through like clockwork in every election cycle?

See my prior post. I will track the corporate Bush shills against the independent pollsters (Zogby, Harris, ARG, Pew, AP) right up to the election. There iss a statistically significant deviation in average poll numbers between the independent pollsters and corporate media pollsters.

Look at ABC, those arrogant Bushbots who created the lie-filled "Path to 9/11". Since 2001, their average Bush monthly rating is 5% higher then independents. That's 68 observations. So what are the odds that the 5% difference is due to random error? Infinitesimal.

The MoE for a 68,000 n-sample is 0.38% with ZERO cluster effect! So the 5% average deviation is THIRTEEN times the MoE (or TWENTY FOUR standard deviations) from the independent average. Quite significant.

It's all in the weightings. Just check Gallup's internals; they always weight republicans higher than democrats. They did it in 2000, they did it in 2004, they are doing it now.

Forty-four percent approval? Right.


John Salmon:

Wow-TruthisAll isn't kidding. That opening, "Alan, You are so right", had me fooled for a second.

Every election that the Republicans win is stolen, I see. And no one supports Bush-well, not in his crowd, anyway. He's (she's) gotten his talking points from DailyKos pretty well memorized...which is a lot of work. There's a great site for hysterical Bush-haters like TIA:


Unlike our friend TIA, Liberal Larry IS kidding.



Bush is at 38% in the Harris poll from 9/15:

"Among registered voters, 35 percent say they are going to vote for the Republican candidate, while 45 percent say they will vote for the Democratic candidate. The race is similar if we look at likely voters, (i.e., registered voters who say they are very or somewhat likely to vote) as 35 percent choose the Republican candidate, while 46 percent say they will vote for a Democrat. Thirty-seven percent of interested voters (i.e., registered voters who say they are very or somewhat interested in the upcoming election) say they are voting Republican while 47 percent say they are voting Democratic."

Color me unconvinced by Gallup.



John Salmon, I post at Daily Kos, and and your characterization of the site is wrong. Just saying, since you brought it up.


Steve Singiser:

Departing from the "I Hate Liberals/I Hate You Back" nonsense...I think Alan is right.

There is NOTHING in this year's polling or voting patterns to suggest that Republicans are so much more motivated to vote to account for a NINE-point gap between registered voters and likely voters.

Indeed, I have begun a study of primary turnout for 2006: in seven of the eight states I have studied to date, Democrats made up a greater percentage of the two-party turnout this year as opposed to the previous midterm (2002).

Also, a micro-analysis of individual races has seen the Republican performance slipping in several key races, rather than surging.

I find it funny how Republican friends of mine completely pooh-poohed generic ballots as useless when they had Republicans down 15 points, but now suddenly find them infallible when they are the only evidence of a tightening election.


Steve Singiser:


Good to see you here...

You raise another interesting point. Several pollsters this year, when tightening their screen to likely voters, have seen the Democratic lead in the generic ballot actually INCREASE.

Harris had that...so did Pew, if memory serves, as well as Rasmussen.



Good point, Steve. I think the advice to wait and see is the best advice. The polling from AP-Ipsos, Pew and harris don't agree with gallup, and Rasmussen suggests that Gallup is right about a bounce but that it's short-lived.

Apologies for I hate/I hate you too, if that's what it sounded like.



I think this may be the high water mark caused by the wall-to-wall 5 year 9/11 anniversary coverage. Barring an attack (which since Al Queda seems intent on supporting Bush's alienating policies, could happen) I believe approval rates will be back under 40 and Congressional preference back to +10 within a few weeks.


This recent polling data tells me that voters have a clear perspective on the war in Iraq...perhaps more cogent than either Party. They feel it is being handled poorly, they know what a civil war looks like, they believe Congress has failed to do its part in guiding and overseeing the executive branch, and they realize that the notion of exporting democracy to the Middle East is a Bush Doctrine that fails to recognize the realities in the region. Finally, they believe that Middle East stability is important and that a withdrawal that leaves Iraq in chaos may well be detrimental to the United States.

That, my friends, is one spot on analysis and suggests that voters have discerned fact from fiction with an impressive demonstration of acuity. Perhaps both parties will someday learn that the truth is, in the final analysis, the most powerful campaign strategy available. Don't hold your breath.

Read more here:



Brent (R) Iraq Vet:

The Dems spend far too much time reading charts and graphs and trying to find out which way the wind is blowing and hating Bush. I haven't seen or heard anyone in the Dem party that is coming up with any real answers or ideas just a lot of whining and crying about how the Reps stole their milk money. The Dems are so consumed with Bush that they have yet to realize that the presidental election is only 2 years away and neither Bush nor Cheney are running but that is who the Dems keep trying to run against. Bush may not have handled the war in Iraq perfectly but if my history serves me correctly when the Democratic's savior Roosevelt was conducting a war American's were intered in camps as threats to national security, products rationed, agencies opened mail and blocked out any references that might give information away and the media was massaged and controlled to monitor what news got out. Bush is bombarded from all sides by Libs that want to tie the hands of the military and the government that are trying to win this war. I have been to Iraq and every Iraqi I spoke to were glad we were there, wanted to enjoy the freedom like we have, and were happy Saddam was gone.


Elizabeth Liddle:

Hi, TruthIsAll!

You write:

"The MoE for a 68,000 n-sample is 0.38% with ZERO cluster effect! So the 5% average deviation is THIRTEEN times the MoE (or TWENTY FOUR standard deviations) from the independent average. Quite significant."

You can't pool the respondents from 68 polls, because you can't ignore the dates. You need to do a time series analysis. I did so, and found, as you found, that ABC tended to shoot high (but not as far above the mean as Zogby was shooting low), although it was only borderline "significant" (p something between .1 and .05). There was, of course, a significant difference between ABC and Zogby.

But the important point is that what these analyses demonstrate is that some of the between-pollster variance is not sampling error but due to methodological differences, as you rightly imply. In other words, at least some of the polls must have "bias". However, what you can't do is use any kind of probability estimate to decide which are the biased poll, although obviously you are entitled to form an opinion based on the methodology.

What you can do, however, is to look at within-pollster trends - and I'm afraid it looks to me as though there is a clear uptick, regardless of each pollster's baseline.


Of course I hope it's just a blip.


Bruce Moomaw:

I'm very suspicious of that Gallup Poll, too, and I don't think this reflects wishful thinking on my part -- it doesn't match most of the others. The only one that has recently shown a similar upward jump is the Fox poll -- about half a dozen other recent ones show nothing of the sort. In particular, see Rasmussen's rolling daily poll results ( http://www.rasmussenreports.com/Bush_Job_Approval.htm ). Whatever you think of Rasmussen's techniques --in particular, as they themselves noted recently, they phrase questions about voter approval of a politician in a way which consistently shows more support for that politician than most other pollsters do -- they seem to monitor CHANGES in voter sympathy pretty well, and what they've shown in Bush's case is a mild upswing to about 55-45 negative during the week after his speech, following which he's slumped right back down to the 58-42 level he was at before (and that's WITH Rasmussen's positive-oriented type of approval question).

Nor are any of the individual polls on state races that I'm seeing anywhere in the last two days -- from Rasmussen, Survey USA, the very reliable Mason-Dixon and Quinnipiac, or anywhere else -- showing anything like the sudden surge that Gallup and Fox have reported. I think those two polls just caught a statistical hiccup -- something which Gallup has been weirdly subject to for some years now, as Alan Abramowitz points out. (Last week they showed Bush much WEAKER than any of the other pollsters did! Which, whatever else it proves, doesn't exactly indicate that they're a Bush shill.)

Parenthetically, Rasmussen -- which yesterday reported pretty dramatic surges for the Dems in several important Senate races -- has just also reported that Lamont is again breathing down Lieberman's neck, within 2 points of him. This will no doubt set off firework celebrations at Daily Kos, although I'm more concerned with those other races -- including their results later today for Maryland. (Rasmussen also confirms that -- if Gov. Perry of Texas had only TWO opponents running against him, instead of three -- there's a real chance he might lose.)


I have just implemented some major updates to my sample-weighting website, which I first introduced during the '04 campaign.

For those of you not familiar with my page, it addresses the issue of what (if anything) to do when the partisan composition of respondents in a poll (i.e., X% Democrats, Y% Republicans) deviates substantially from the proportions we would expect from large numbers of previous polls.

From this point forward in the '06 campaign, I will chart the partisan composition of the sample, Bush approval rating, and Generic Congresssional Ballot reading from every national poll, provided the necessary data are available.

The page is located at:



fat karl:

With all due respect Brent, then why were they trying to kill you? I don't think many Iraqi's are glad we are there.

I have been to Iraq and every Iraqi I spoke to were glad we were there, wanted to enjoy the freedom like we have, and were happy Saddam was gone.


Bruce Moomaw:

With nice timing, Survey USA has just come out with its latest nationwide poll of Bush's popularity -- still 59-39 negative
(http://www.surveyusa.com/50State2006/50StateBushApproval060919Net.htm )
This -- along with another half-dozen recent polls -- provides further
evidence that Gallup's supposed sudden surge for him is a statistical fluke (like its poll last week showing him much WEAKER
than all the other nationwide polls did).

Also, Survey USA shows yet another GOP Senate seat suddenly turning wobbly -- namely, Arizona (which the Dems had just about given up on): http://www.surveyusa.com/client/PollReportPopup.aspx?g=7bc891cf-4d78-4a76-981e-f7769f103f36&q=31417 .


S.M. Stirling:

Jeeze, talk about people throwing rocks at the messenger...

The fact of the matter is that there's a strong trend towards the GOP and we're in denial. We peaked in July-August.

The 48-48 split is worse than it seems; that LV poll from Gallup always overstates the Democratic numbers, usually by 5-8%.

So we're actually down.

I don't see why this is a surprise. We haven't hit 51% in a national election since 1964.


S.M. Stirling,

The Dems have won at least 50% majority in the last 4 presidential elections. Just remove third parties from the vote.

1. 1992
Clinton 43% (43/(43+38)= 53.1%
Bush 38%
Perot 19%

2. 1996
Clinton 49% (49/(49+41)= 54.4%
Dole 41%
Perot 9%

3. 2000
Gore 51.0mm (51/(51+50.5)= 50.2%
Bush 50.5mm
Nader 3.5mm

4. 2004 (this one will surprise you)

Kerry won the 12:22am National Exit Poll (NEP) by 51-48%. In the 2:05pm
Final NEP, weights AND vote shares were adjusted to MATCH the vote count.
Bush won the Final, 51-48%. That the Final NEP 43% Bush/37% Gore weights
are MATHEMATICALLY IMPOSSIBLE is beyond dispute.

Change to PLAUSIBLE weights in the Final and Kerry is the 51-48% winner -
BETTER than 51-48%, since his 12:22AM vote shares were SUBSTANTIALLY
REDUCED in the Final.

To believe the Final NEP, you must also believe:

1) That 52.57mm (43% of 122.3mm), or 108% of the 48.7mm Bush 2000 voters
still living, turned out in 2004 - FOUR MILLION more Bush 2000 voters than

IF we assume a PLAUSIBLE 95% Bush 2000 voter turnout (46.25mm, or 37.83% of
122.3mm), THEN the Final NEP OVERSTATED Bush voter turnout by 6.3 MILLION.

2) On the other hand, the NEP apparently UNDERSTATED Gore 2000 voter
turnout. The 37% weighting means that 45.24mm voted in 2004, or JUST 92% of
the 49.2mm still living. So 8% stayed home? Not likely.

IF we assume PLAUSIBLE 95% Gore 2000 voter turnout (46.75mm or 38.24% of
122.3mm), THEN the Final NEP UNDERSTATED Gore voter turnout by 1.5 MILLION.



(Impossible 43%/37% weights)

BUSH WINS: 62.5-59.3mm (51.1%-48.5%)

Voted..2004............Vote Share..........Votes (mm)
2000 Votes Weight Kerry Bush Other Kerry Bush Other
No 20.79 17% 54% 45% 1% 11.22 9.35 0.21
Gore 45.24 37% 90% 10% 0% 40.72 4.52 0.00
Bush 52.57 43% 9% 91% 0% 4.73 47.84 0.00
Nader 3.67 3% 71% 21% 8% 2.60 0.77 0.29

Total 122.27 100% 48.48% 51.11% 0.41% 59.28 62.49 0.50



(Adjusted for plausible weights)

MAXIMUM WEIGHTS (100% turnout): 40.25% Gore/39.82% Bush
PLAUSIBLE WEIGHTS (95% turnout): 38.24% Gore/37.83% Bush
NO CHANGE in vote shares.

KERRY WINS: 62.6-59.2mm (51.2%-48.4%)

Voted..2004............Vote Share...........Votes (mm)
2000 Votes Weight Kerry Bush Other Kerry Bush Other
No 26.22 21.44% 54% 45% 1% 14.16 11.80 0.26
Gore 46.75 38.24% 90% 10% 0% 42.08 4.68 0.00
Bush 46.25 37.83% 9% 91% 0% 4.16 42.09 0.00
Nader 3.04 2.49% 71% 21% 8% 2.16 0.64 0.24

Total 122.27 100% 51.17% 48.42% 0.41% 62.56 59.20 0.51



Elizabeth Liddle:

OK, looks like a 9/11 anniversary blip. Any more blip producers on the horizon?


Gary Kilbride:

Here's a basic reference point for Bush's approval rating, statewide Strategic Vision polls in Michigan and Wisconsin compared to similar polls by the same firm a month earlier. He is up 4% in Michigan and 5% in Wisconsin. Those are blue states that will trend lower than the national average but it's remarkable rubbish to deny there has been a bump, or pretend it won't continue as gas prices plunge.

And it's the reason I insisted all year Democrats needed to prioritize their own favorable ratings, and put something simple and memorable into the national vocabulary. Now the party is stuck with a vote-against dependency, which will never magnet as many followers or have as much landslide potential as a populous determined to storm the polls to vote FOR a person or philosophy.

Also, note the approval ratings on the economy, an economy which Bush continues to rave about, calling it great and saying he expects November will be a referendum on the economy. Meanwhile, Strategic Vision state polls consistently indicate Bush's approval number on the economy are horrid, lower than on Iraq. Wisconsin is the only state SV is polling in which the approval rating on Iraq is slightly higher than on the economy.


Michigan: (August 25-27)

Overall: 34%
Terrorism: 49%
Iraq: 35%
Economy: 27%

Michigan: (September 15-17)

Overall: 38%
Terrorism: 52%
Iraq: 39%
Economy: 28%

Wisconsin (August 11-13)

Overall: 32%
Terrorism: 43%
Iraq: 22%
Economy: 26%

Wisconsin: (September 15-17)

Overall: 37%
Terrorism: 49%
Iraq: 25%
Economy: 29%



If you remove the agendas from the analysis (I told you it's the economy, stupid; it's about the stolen election, etc), Lizzie's right. it's a small blip as measured by prof. franklin, and rasmussen, and other recent polls (Harris, SUSA, Pew) other than Gallup say it aint much and isn't lasting. SM Sterling's analysis misses the mark. There's no strong GOP trend other than Gallup. What there is can be seen as shaky R's susceptible to persuasion and good faith in the WH returning from a dismal summer low. IOW, a combo of gas prices and the 9/11 anniversary. That's a weak and not a strong GOP trend as of this date.

I look forward to Prof. Franklin's continued aggregate to see this more clearly. Based on previous performance, Gallup is important but hardly definitive.



I think I can explain this data fairly easily-

it is a function of the frequency of the polling.

If we look at Rasmussen which polls every day on presidential approval we see that just after 9/11 there was a small bounce up and then back down in the president's approval ratings. see here:


approval started at 41-41 before hand, got up as high as 47 on the 14th and came back down to 41-42 by the 17th.

What was the data of the Survey USA poll? sept 15th-17th. Basically the back half of the bounce. Rasmussen of course is far from infallible but the SUSA result meshes perfectly with what Ras is getting once we account for the difference in sampling rate.



whoops where I say SurveyUSA above read Gallup. My bad.


Elizabeth Liddle:


Well, that's sort of what I was getting at in my earlier post. Given that short term blips may be real, you are going to have between-pollster variance that's just due to variance in exactly when, relative to a blip, the data was collected. So it's only daily polls that have a sampling frequency capable of detecting high frequency blips. What's signal to Rasmussen is noise to Gallup.





yes, elizabeth, there is a blip...or more accurately, a steady drip-drip-drip. October Crude Oil today on the NYMEX hit $60.60 a barrel, a six month low; it's come down $17 from the peak.



You're right gjdodger. Once again, the Republicans are maniupulating the prices of gas and oil just in time for the election.



Well I know now we will keep both houses because the democrats are already shouting "Stolen elections". Do they know how stupid they sound?


David :

You democrats have been brainwashed by the likes of moveon.org, michael Moore, etc. You are all sick.



Right. The drop in oil prices has nothing to do with the end of the driving season, the lack of any threatening hurricanes, El Nino forming in the Pacific (predicting a mild winter) and the massive investment in oil drilling and refining sparked by the high prices of earlier this year. It's all a Republican plot, aided, of course, by their good "friends" in Iran, Venezuela and Russia who primarily guide the price of oil.



You talkin' to me?
(said with DiNiro accent)

Brainwashed? Stupid? Stolen elections?
These terms are all associated with Bush.

How about addressing my math?
And stop the whining.

You remind me of what is said about TIA on FreeRepublic and the Conservative Underground.


Brent (R) Iraq Vet:

Fat kar, have you ever been there? Have you spoken with run of the mill Iraqis? Have you seen the thankful expressions and smiles on the faces of the people when the new water plant we helped pay for comes on line, or a new school opens. The problem with Libs is they rely on a liberal media to give them all their information and I can say from first hand experience that the media tells the stories they want told. The hundreds or even thousands of improvement projects that are going on over there are rarely, if ever, covered because that would show progress and the media can't show that they must show the one car bomb. The insurgency makes up less than 30% of the total population and judging by the number of foreign fighters caught most of that number are not Iraqi. If you look at the high profile fighters we have caught or killed most are from somewhere else other than Iraq. Iraqis are happy that we have brought them freedom but it will be a struggle to to make the process work. We have been working on it for over 200 years and I think sometimes we still have work to do.


Watch the divergence in the average Bush rating between independent and corporate pollsters widen as we get closer to the election.

Divergence in Average Bush Ratings
Independents:Pew, Harris, AP/Ipsos,Zogby,ARG
Corporates: Newsweek, Time, Fox, CNN, CBS, ABC, Time, NBC

Ind 52.24
Corp 55.98
Diff 3.74

Latest 12mo:
Ind 36.87
Corp 38.50
Diff 1.63

Sept. 2006:
Ind 37.75
Corp 40.40
Diff 2.65

MoE Group NWK Fox CNN Pew Harris CBS ABC Time NBC AP Zogby ARG
1.58 Indep - - - 37 38 - - - - 39 37 na Avg 37.75
1.41 Corp na 40 41 - - 37 42 na 42 - - na Avg 40.40
Diff 2.65

0.05 All 55.46 56.01 56.58 51.83 54.12 54.97 57.95 56.58 54.29 51.00 52.00 na Avg 54.74

0.19 Indep - - - 51.83 54.12 - - - - 51.00 52.00 na Avg 52.24
0.14 Corp 55.46 56.01 56.58 - - 54.97 57.95 56.58 54.29 - - na Avg 55.98
Diff 3.74

Latest 12 mo
0.17 Indep - - - 36.85 36.15 - - - - 37.50 37.58 36.25 Avg 36.87
0.14 Corp 36.88 38.69 39.17 - - 36.31 40.09 39.78 38.60 - - - Avg 38.50
Diff 1.63



say what you want but polls are what run the country. w/out them we might not get our canidates that we want to run really run.Fact
most surveys are run by pollsters they sell their data to the cannidates so what the man running for whatever can see the results and that could tell him to run his platform on something else. Would you throw your right to vote away. you are when you dont respond to a survey.So be an american an vote with ur pollster


Post a comment

Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.