The O’Leary Report in conjunction with Zogby International and Southern Methodist University’s John Goodwin Tower Center conducted a survey from December 15-19, 2003 of 1,200 voters nationally with a margin of error of + 2.9%. The first installment of the Zogby Values poll for 2004 is unique in that we provide a breakout analysis of voter responses in states won by George W. Bush in 2000, hereafter known as “red states,” and voters in states won by Al Gore or “blue states.”
Interesting survey I thought y'all might enjoy.
Posts: 13123 | Registered: Feb 2002
| IP: Logged |
posted
Wow. The first thing that stood out, to me, is the remarkably odious use of loaded terminology to describe the positions.
The bias here is actually physically PAINFUL. On reviewing the actual questions, I've decided that I'm going to use this survey as an example of how you can write a loaded survey to get the results you want.
posted
Here's an example, from their analysis of an abortion question:
"Partial-birth abortion and parental notification were not tested in this survey as those numbers have risen to acceptable levels in the pro-choice precincts and so we determined not to waste a poll question on them."
I think it's pretty clear that any poll which looks for "acceptable levels" of a given response is biased. Posts: 37449 | Registered: May 1999
| IP: Logged |
posted
I also agree that the questioning was poorly worded. For instance, the "Was the 2000 election stolen or legitimate?" question. I have questions of legitimacy, but I don't think it was stolen, deliberately or malevolently.
posted
I still think the survey is interesting. Is it 'wrong'? I don't know.
I think it's a little early to cry bias unless we can show that the survey is trying to reach certain conclusions, though I do agree the wording is poor.
Posts: 13123 | Registered: Feb 2002
| IP: Logged |
posted
I don't think it was POORLY worded. I think it was worded very well, since its deliberate goal was obviously to obtain the kind of answers they wanted. Consider, just as an example, the very first question, which is whether people approve of the decisions both Clintons made -- financially, politically, and morally.
Think about this question for a second.
First off, it's the first question of a political poll that's about general attitudes on current politics. So why the Clintons? And why first? Two big reasons, IMO, is to get the person taking the poll to start thinking about the Clintons and getting upset about them, mentally linking those positions back in the late '90s to liberal positions today (as reflected in the rest of the poll). It also, IMO, seeks to remind people that Hillary Clinton is still around, and tie her yet again to her husband.
Now look at the wording. Many people I know approve of the job Clinton did in office, and regard him as a reasonably effective president; relatively few of them, however, approve of his moral decisions. I do not think it's an accident that this is a SINGLE question, rather than a set of three questions about approval of each facet of his performance.
Let's look at another question: "Should we help create a stable government in Iraq, even if we have to sacrifice a bit, or should we hand Iraq over to Russia (old enemies), France (snarky frogs), Germany (the Nazis!), China (the Red Menace), and NATO (whatever that stands for)?"
Now let's look at their question about religion:
quote: Thinking about the presidency, which of the following are important to you? 1) Having a president who believes in God and is deeply religious, but whom most Americans believe has done a good job managing the economy and foreign policy or 2) Having a president who is not religious, but whom most Americans believe has done a good job managing the economy and foreign policy.
See the amusing bit here? On one hand, you can pick a deeply religious president who's done a good job running the country, OR you can pick one who ISN'T religious, but has done a good job running the country. Now pick which one is "more important." We're not talking Sophie's Choice, here; it's not even TASTER'S Choice.
Now look at this one:
quote: Do you agree or disagree that American firearm manufacturers who sell a legal product that is not defective should be allowed to be sued if a criminal uses their products in a crime?
Look at the wording. It's a "legal product" that "is not defective." Again, this is leading the witness.
And finally, here's a question which, while not the last one, is the last one I'll address, because it says it all:
quote: Suppose you earned one million dollars in your job. Which of the following tax rates do you think would be fair to you and the federal government? Less than 10%, 10-20%, 21-30%, or more than 30%.
Anyone who's ever constructed a poll that contains ranges knows that people are reluctant to EVER select numbers at either end of the range, unless they're ideologically committed to one specific "high" or "low" mindset. By deliberately setting the tax rate up in this way, the pollsters are ignoring the fact that the current 30%+ income tax on the rich -- which, by human nature when taking polls alone, is going to wind up "unfair" on this poll -- is not historically high. Had they included some of the historical levels, the range would have reached up to 90% or so -- and I'd wager that more people would have been willing to go as high as 50%. Of course, this discussion of tax rates also oversimplifies taxes, implying that this is the actual percentage of income paid, rather than a bracket; someone who hits the 38% tax bracket does NOT pay 38% of his income in taxes, but rather has the portion of his income that exceeds that bracket taxed at 38%. While the Beatles might sing about the taxman taking 19 and leaving them 1, that reflects a certain misunderstanding of the way tax brackets work.
quote:I think it's a little early to cry bias unless we can show that the survey is trying to reach certain conclusions, though I do agree the wording is poor.
It doesn't really matter what the intention is; the point is that the way it is worded biases respondents. A broken survey is a broken survey no matter what its intentions are.
Posts: 2443 | Registered: Apr 2002
| IP: Logged |
Printer-friendly view of this topic
![[Smile]](smile.gif)
