The latest surveys from Gallup/CNN/USAToday and the Washington Post/ABC News were released last night that collectively take the most in-depth look yet at public opinion at the response to Hurricane Katrina. No quick blog post could do justice to all the data available in both surveys. I’ll post the links and some of topline results we have been focusing on most closely now and do some cross-survey comparisons later in the day.
As usual, analysis of the Post/ABC poll is available in both a front page Washington Post story and an online summary from ABC News. The Post also has the complete results and demographics available by online PDF. The link to ABC’s complete PDF version was not functioning as of this writing is now functioning fine.
The key results for the Post/ABC poll of 1,201 adults, conducted Tbursday through Sunday last week (9/8-11):
- Overall Bush job rating: 42% approve, 57% disapprove, a new low, down from 45%-43% approve-disapprove just two weeks ago.
- Rating of Bush on "the situation caused by Hurricane Katrina:" 44% approve, 54% disapprove, down from 46%-47% approve disapprove two weeks ago.
- Rating of the "federal government’s overall response to the situation caused by Hurricane Katrina:" 38% excellent or good, 62% not so good or poor, down from 48% excellent/good, 41% not so good/poor two weeks ago.
- 50% say Bush as a "strong leader," down from 62% in May.
- 49% say Bush be "trusted in a crisis," down from 60% in May.
- 76% support "an investigation by an independent commission like the one that investigated the 9/11 attacks."
The analysis by ABC’s Gary Langer emphasized the intensity of feeling in the Bush job rating:
As striking as Bush’s rating – his disapproval is higher than the worst for either of his last two two-term predecessors – is the intensity of sentiment against him: Forty-five percent of Americans "strongly" criticize Bush’s performance in office, an unusually deep well of disapproval. Far fewer, 27 percent, strongly approve.
As of this hour, the Gallup/CNN/USAToday results are available online only in a CNN summary and a story by USAToday‘s Susan Page. Separately, USAToday provides the complete text of each question, but with results broken out by race only, omitting the overall results. Presumably, Gallup will post its own summary and more complete results on its website later today (UPDATE: it is now online here).
So where is the "official first George W. Bush job approval rating after the Hurricane" that Gallup’s Editor-in-Chief Frank Newport promised and speculated about last week? One will not find it in the CNN online summary, which oddly omits discussion of the overall Bush job rating altogether (as did Suzanne Malveaux’s quick on-air presentation of the poll results at 7:08 am EST). It does appear in the USA Today article, although the placement (paragraph 10) and Page’s description may provide a clue to CNN’s omission of the overall job rating form their online summary:
Overall, the president’s job-approval rating is 46%, essentially the same as the 45% rating in the Gallup Poll 10 days earlier.
That’s at odds with record-low ratings of 38% in a Newsweek poll and 39% in an AP-Ipsos poll, both released on Saturday. An ABC News-Washington Post survey released Monday put Bush’s approval rating at a record low 42%.
One possible reason for the discrepancy: The USA TODAY survey was taken later than the Newsweek and AP polls, though the timing coincided with the ABC poll.
Other results from the Gallup survey, conducted among 1,005 adults conducted Thursday through Sunday last week (9/8-11):
- Rating of Bush’s "handling of the crisis" (from the CNN summary): 43% approve, 54% disapprove.
- Is Bush a "strong and decisive leader?: 52% agree, down from 60% just two weeks ago.
- 70% favor "an investigation by an independent panel into the problems with the government’s response to the hurricane."
Again, these surveys go into much greater depth on reactions to Katrina, especially with respect to the performance of federal, state and local officials and on perceptions of racial disparities. The summaries are worth reading in full. I’ll update this post with the Gallup.com links when they become available. UPDATE: The full summary by Gallup’s David Moore is now up and should remain available to non-subscribers for at least 24 hours.
Mark – Check out our post on this regarding the “oversampling” of blacks and the Post’s not publishign that fact in their later story. Check it out here w/ appropriate links:
http://www.anklebitingpundits.com/index.php?name=News&file=article&sid=2293&mode=nested&order=1&thold=0
The Post does address the oversampling of blacks and subsequent adjustment in their demographic results table(which were not available until later today, after we posted the above story).
But I find it strange that unlike other polls they publish, the Post doesn’t give data on Party ID, age, income, etc. Wonder why that is.
To those of you who won’t bother with the http://www.anklebitingpundits.com link–they published the first (Sept. 12th, 5:31 p.m.) Washington Post writeup of this poll. In that writeup, the WaPo admits that its poll of “1,200 randomly selected adults” included “an oversample of 200 African-Americans”. Curiously, the Sept. 13th WaPo writeup is identical, except that it censors all reference to that “oversample of African-Americans”. If the WaPo can’t be honest about their own poll, why should it be accorded any credibility? [Note I’m not challenging the credibility of the poll–though I might do that in another post–but rather the WaPo deceit about how the poll was taken]
Put more basically, if the WaPo deliberately oversampled Blacks, then the claim that they polled “1,200 randomly selected adults” is a lie. Alternatively, if their random sample just happened to come up with 16.66% Blacks (a 50% higher proportion than in the national population), then their random sampling methods are a joke.
Now that the complete pdf version is available for reading, I’ve found yet another Post lie. Where the initial article (afterwards censored) claimed the oversample was of “200” Blacks, the pdf says the oversample was “271” Blacks.
The poll contains some interesting results as to the public’s disgust a state and city relief efforts. This being the Post, as expected these results (which would blame Democrats) weren’t in the article.
As to slanted questions in the poll, take a gander at, among others, question 26. It asks if New Orleans were less Black, do you believe either a) the government would have responded faster, or b) the response time would have been the same. Note the framing of the available choices–the government response was either anti-Black, or race neutral. No mention of the logical option c), that the response time would have been slower, an especially odd omission since poll questions generally give three options–more of something, the same, or less of something.
The “what keeps John Kerry and Al Gore up at night
Apparently, the Administration’s handling of federal budget is indicative of their ability to manage political capital. Zing!
Civilwarguy, you are mistaken.
Here is what the ABC pdf actually says: “This ABC News/Washington Post poll was conducted by telephone Sept. 8-11, 2005, among a random national sample of 1,201 adults, including an oversample of blacks for a total of 271 black respondents. The results have a three point error margin for the full sample, six points for blacks.”
Here is what the WP pdf actually says: “The latest Washington Post-ABC News poll is based on telephone interviews with 1,201 randomly selected adults nationwide, and was conducted Sept. 8-11, 2005. The poll included an oversample of 200 African American respondents. Final results were adjusted to reflect the country’s actual racial distribution.”
Translation: A random sample of 1001 adults was conducted. There were 71 African American respondents. Then a random sample of 200 African Americans was conducted. Results were then weighted to reflect US demographics (74% white, 11% black, etc.)
So, there’s no discrepancy in the oversample numbers, and it’s still a random sample.
The oversampling is used to reduce the sampling error in small groups if results are to broken out by group. If they had not oversampled, the sampling error from just 71 African Americans would have been about +/- 12%.
[Things would be simpler if all the writers described the methodolgy exactly the same. But that’s not likely to happen.]
Nor is the Post “lying” as CivGuy claims on his blog. The black oversample was also “randomly” sampled (thus the whole sample is accurately described as “random”).
Don’t shoot the messenger. The trend for Bush’s approval rating is clear.
Paul, Lynxx, what you have is NOT “a random sample of 1,201 adults”, as the WaPo claimed. What they did was TWO samples, one the random sample of 1001 people, the other the “oversample” of 200 Blacks. The 1,201 figure is the product of TWO samples, not ONE, with one of the samples not being random. I stand by my statement.