Polls Do they mean anything?

Excellent article Dan.

The point is that people are most often willing to trade off the strict adherence to a principle for practical considerations.

So very wrong, Dan.

The ABC News/WaPo poll is based on a false premise--not the fact Bush's name wasn't used. The false premise is that the respondent has two choices: 1. Violate privacy and be safe from terrorists or 2. Respect privacy and become a victim of terrorism.

This is what's known in statistical inference as framing bias which drives a respondent to select between two choices--one or more of which are false.

As we've seen in the recent Bush illegality concerning FISA--obtaining a warrant has never prevented the necessary intelligence to prevent terrorism. In fact, the warrant could be obtained after surveillance/eavedropping had begun. Thus, the implication that obtaining a warrant somehow limits intelligence into terrorism is false.

This is more of the same from the conservatives: polls aren't to be trusted because Bush's numbers are abysmally low. You can be assured, however, that if Bush's poll numbers were in the 60s or higher--we'd be hearing these same critics touting the polling companies.

What conservatives tend to forget is that a polling company's accuracy is what its reputation is based on. If they're fudging results or conducting flawed polling--they're not going to stay in business.

And if these polling companies are so biased against conservatives--why is it the GOP spends so much money on these polling firms? Why is Bush responding to these "biased" and erroneous polls?

The answer is apparent, of course.

As we've seen in the recent Bush illegality concerning FISA--obtaining a warrant has never prevented the necessary intelligence to prevent terrorism.

I feel secure now that I have your word on that.

but we're to believe that these wiretaps are preventing terrorist attacks, right greg?

Jade, Interesting comment:

So very wrong, Dan. The ABC News/WaPo poll is based on a false premise--not the fact Bush's name wasn't used. The false premise is that the respondent has two choices: 1. Violate privacy and be safe from terrorists or 2. Respect privacy and become a victim of terrorism.

 I am 'very wrong' because I said the poll didn't ask the right questions:

But look at the question. It doesn't even come close to issue under debate.

But you are right because you said that the poll did not ask the right quest.........hmmm.

And how in the world did you come to the conclusion that I thought it was wrong because Bush's name wasn't used?

Jade, while I understand the impulse to argue since I too suffer from it, I would ask you to please read carefully before accusing me ( or anyone else) of making false statements.

It is true that surveillance can be conducted without a court order pursuant to Title 50, Chapter 36, Subchapter 1, Subsection 1802 of the FISA of 1978. Text as follows:


Release date: 2005-03-17

(a)
(1) Notwithstanding any other law, the President, through the Attorney General, may authorize electronic surveillance without a court order under this subchapter to acquire foreign intelligence information for periods of up to one year if the Attorney General certifies in writing under oath that—


(A) the electronic surveillance is solely directed at—


(i) the acquisition of the contents of communications transmitted by means of communications used exclusively between or among foreign powers, as defined in section 1801 (a)(1), (2), or (3) of this title; or


(ii) the acquisition of technical intelligence, other than the spoken communications of individuals, from property or premises under the open and exclusive control of a foreign power, as defined in section 1801 (a)(1), (2), or (3) of this title;


(B) there is no substantial likelihood that the surveillance will acquire the contents of any communication to which a United States person is a party; and


(C) the proposed minimization procedures with respect to such surveillance meet the definition of minimization procedures under section 1801 (h) of this title; and
if the Attorney General reports such minimization procedures and any changes thereto to the House Permanent Select Committee on Intelligence and the Senate Select Committee on Intelligence at least thirty days prior to their effective date, unless the Attorney General determines immediate action is required and notifies the committees immediately of such minimization procedures and the reason for their becoming effective immediately.

(i) the acquisition of the contents of communications transmitted by means of communications used exclusively between or among FOREIGN powers

(ii) the acquisition of technical intelligence, other than the spoken communications of individuals, from property or premises under the open and exclusive control of a FOREIGN power

(B) there is no substantial likelihood that the surveillance will acquire the contents of any communication to which a UNITED STATES PERSON is a party

~~

thanks, diana.

Dan: You're still wrong; the issue wasn't that the "right questions" weren't asked. The issue has to do with framing bias, which is altogether different matter. And it plays into the bunkum conservatives are spouting about polls being unreliable/dishonest etc.

There really is no right or wrong question that can be asked in a poll. It's the analysis of the response that's important. I'll give you an example--suppose a poll were to ask: Would you like to pay more or less in taxes? Well, I'd guess that 99% of respondents would say 'less.' Does this mean tax cuts are winning issue for a candidate? Probably not because the qpoll question didn't furnish enough info or context. Most people would dearly love to pay less in taxes but they may place other agendas (better schools, roads, etc) above their tax-cutting desires.

I held off commenting on your last comment here Jade because you did not seem to want to see. 'Framing bias' is a matter of asking the wrong questions. It means asking questions designed to elicit responses that appear to support one side or the other of a debate rather than asking questions that would elicit responses about the actual positions of the various sides of the debate. In other words a poll with 'framing bias', deliberately or not, asks the wrong questions.

I'm really fond of John Zogby's polls. He uses a widely demographically distributed sample of households and such.

I know from the 2004 election season, as a Deanie, that whenever there was a poll that impacted the Dean campaign on a media web site -- one of the ones where anyone who comes to the site can vote -- the emails would go out on *every* grassroots Dean campaign list, and thousands of Deanies would rush to a URL for a moment in their day to skew the poll.

I'm sure the same thing happens with every wired community. The Republicans are sending out alerts. The Dems. The Greens. The Chartreuses. Who knows what big and little groups are wired enough today to skew an online poll?

So, I'll stick to the more professionally sampled polls if I even want a wild a$$ed guess what public opinion is, thank you.

Zogby's opinion on his front page today?

By a margin of 52 to 43 percent, citizens want Congress to impeach President Bush if he wiretapped American citizens without a judge's approval.

That opinion was gathered scientifically, not by fanatics of either side mashing radio buttons on a webpage.

Ask a simple question, get an accurate poll. Infuse purposely confusing political aspects, get a crap poll.

Dan,

Nobody answered your initial question: Do polls mean anything? The answer is no, certainly not by themselves. As the comments here indicate, polls are merely tools, or tactics, that both sides of an argument use to bolster their points. Polling is a science. It's NOT reporting, or fact finding, or truth gathering.

Jade has a good point. When asking a question that requires respondents to choose between two items, the items themselves should be mutual exclusive and exhaustive. Otherwise, they result in a bias, just as Jade points out.

As we all know, polls are often conducted to "prove" points of policy, not to investigate the "truth". In comparison, a well designed survey would look at some of the questions, and offer another option such as "Other" allowing the respondent to offer another response. Pre-testing this would allow the researcher to then create a battery of answer items that meet the criteria of mutually exclusive and exhaustive.

(sorry about the lecture, but I'm a marketing researcher, and I can't avoid jumping in on these types of things)

I agree with Jade (for a change, lol!) and Eric. Many polls fail to provide a definitive answer to a particular question. As in the case of CNN-Gallup polls that were conducted during the last election, the media tends to latch on to the results making common but misleading conclusions.

Another problem is that many polls claim to be a "statistical sampling", but that may or may not be the case. While unlikely, it is possible that all the people polled were Bush supporters or all opposed to his administration, which would slant their opinion. A true, scientifically controlled survey should be conducted over and over again to be sure there's a representative sampling of the population.

Donald's comment, I feel, has hit the bullseye, except that polling is NOT a science, it is a tool as he stated. Just by rephrasing a question or a premise in the poll, you can easily garner totally different results. A "poll" is what it is ... something to club you with.