[I think GWB's public support, like Lake Pontchartrain itself, is broad but shallow. In any event, Doug, I thought you were a connoisseur of polling techniques. I'd like to take the liberty of posting the following from the Marxism list:]
Subject: ABC News Pimps for the President with Poll From: Joaquín Bustelo <jbustelo at bellsouth.net> Date: Mon, 5 Sep 2005 09:48:10 -0400
The first poll on Bush's handling of Katrina and the resulting New Orleans flood is out, and to hear the sponsoring news organization tell the story, Bush is doing o.k.
This was a quickie one-day poll with 500 respondents, which ABC News chose to do the evening of September 2, the day that Bush was down visiting the hurricane victims, while favorable, staged images of him filled the airwaves and before anyone had a chance to point out they were faked (such as the photo-op by a levee breach repair operation that showed Bush surrounded by all sorts of heavy equipment, all of which but one piece had vanished by the next morning).
According to the poll write up, people are extremely critical of the government's response, but "far fewer take George W. Bush personally to task for the problems, and public anger about the response is less widespread than some critics would suggest."
But when you look at the numbers, the story is significantly different. Those polled disapproved of the President's handling of the situation 47% to 46%, and of the Federal Government's handling of it by 51% to 48%.
Notice I say "those polled," which is different from what ABC News asserts: "Forty-six percent of Americans approve of Bush's handling of the crisis," etc. The reason is that there is every reason to be suspicious of such a quickie poll, especially one that waits until the last paragraph of the fourth jump to give only a couple of bare details of its methodology.
The poll was done of 500 people, and supposedly had a margin of error of 4%. Statistical science says it ain't so. According to the sample size calculator here: <http://www.isixsigma.com/offsite.asp?A=Fr&Url=http://www.raosoft.com/sa mplesize.html>, the real margin of error is 4.34% at a 95% confidence level. Of course, ABC News doesn't say what the confidence level is.
In addition, there is a rounding error introduced by ABC News and its pollsters of 0.5%, which makes the overall error 5% or as close to as makes no difference. Assuming, of course, a perfect sample. Which this could hardly have been.
A one-night --especially Friday night-- poll with no follow up on successive nights is likely to give you a skewed sample. You're basically selecting for a less active, less social, older, layer of the population. That's on top of the increasing problems with telephone polling, including millions of people who no longer have land telephone lines, and the tens of millions who simply refuse to cooperate when a pollster reaches them.
But the most egregious problem with the poll isn't its methodology, bad as it was, but its interpretation and lack of context.
For cover, ABC News does allow that Bush's rating on Katrina "compares poorly with Bush's 91 percent approval rating for his performance in the immediate aftermath of the 9/11 terrorist attacks, but it's far from the broad discontent expressed by critics of the initial days of the hurricane response."
However, the article fails to draw out and make explicit the point that you almost always get a "rallying around the flag" effect that benefits whatever nincompoop happens to be in the White House in the immediate aftermath of a crisis or catastrophe. Viewed in this light, and taking into account that the poll was done on the day --the only day-- when Bush succeeded in projecting himself favorably in the media, the assertion that the poll's results are "far from the broad discontent expressed by critics" is more spin from Karl Rove's desk than journalism.
Especially when you take into account that more than two thirds of those polled say the federal government was unprepared (and three fourths say the same thing about the state and local government, which is the line the white housed and conservative news cabal have been pushing).
Perhaps the most damning confirmation that ABC News used the poll to pimp for Bush comes from the polls print media co-sponsor, the Washington Post, which presented the results very, very differently.
The Washington Post led with a result which was unassailable even given a smallish and possibly flawed sample: "72% Say Gas Scalping Is Tied to Storm -- Majority in Poll Blame Gouging on Government's Handling of Price Surge."
The Post also presents the results in relation to Bush, but much more cautiously: "Slightly less than half -- 46 percent -- approve of the way Bush has handled relief efforts while 47 percent disapprove, a result that might offer some cheer to beleaguered White House staffers who feared a stronger negative reaction."
It also is careful to indicate the moment when the poll was carried out in a week of tumultuous events: "A total of 501 randomly selected adults were interviewed Friday night after Bush visited the Gulf Coast region and as National Guard troops, emergency supplies and relief workers began moving into the stricken city of New Orleans."
In addition, the Post follows the laudable practice of making fuller poll results available in tabular form with the exact questions and answers. From that we learn, for example, that those who described themselves as "strongly" approving of the President's handling of the situation was barely 26% against 31% who strongly disapproved; and on various specific issues those polled gave the following excellent/poor numbers (the two extremes) for the Fed's response:
looting and crime 4%/37% delivering food, water and medical help 10%/29% search and rescue 19%/29% oil supply and gasoline prices 3%/50% evacuating, resettling people who lost homes 8%/29%
Overall, the negative opinions on each specific item save search and rescue were much more strongly negative than the positive opinions. And whereas on Bush's overall evaluation 8% had a neutral or no opinion, on the specific items those who had neutral or no opinion averaged less than 3%.
All of this paints a much more negative picture for Bush than the headline number offered by ABC News.
It is interesting, and perhaps a hopeful sign, is that other news organizations known for having a penchant for quicky polling, such as CNN, haven't released any poll results, one hopes because they realize the folly of phone polling over a labor day weekend.
And the way ABC News used the results of its polls suggests that more was involved than a journalistic desire to get a jump on the competition, as the difference between the way ABC and the Post presented the results of the same poll make clear.
<http://abcnews.go.com/US/HurricaneKatrina/story?id=1094262&page=1>
<http://www.washingtonpost.com/wp-dyn/content/article/2005/09/03/AR20050 90301164.html>
Joaquín
<http://www.columbia.edu/~lnp3/msg71231.html>
Carl