analytical question on polls (for you geniuses)

[historian]

Polls are misleading since the responses will vary according to the question (if the person being asked is honest).
For example if the question was "should bush be re elected?" and the answer was 60% no, the other question "should kerry take bush's place" will also have a high negative since most people dont know kerry, are die hard republicans or any other variety of reasons.
Its really a matter of semantics and the ideology and integrity of the person being interviewed.
Am i right? If not pls say so and ill go massage my ego.

[shraby]

1. I worked in market research

person to person in a mall. Believe me, the questions asked make a whale of a difference. We had one client who monitored some of the responses to his questionaire and was mad. Had to go back to the drawing board and change the questions cause he wasn't getting the "right" responses.

[CharlesGroce]

2. also bear in mind the sampling population

such as when CNN has a poll of it's viewers. Well when they present their findings they say things like '50% of Americans support Bush' when in fact it's '50% of CNN viewers support Bush', a big difference since not all Americans watch CNN, and certainly people who watch CNN are influenced by CNN.

[rock]

3. Polls prepared by a competent scientist or mathematician

can be uncannily accurate. The first step is to define what the question is. To make it as operational as possible, that is objective, even though it's asking an opinion. CNN is notorious for having its questions as wobbly as possible. So you are right. It's a semantic problem first.

[iamjoy]

4. I'm Not A Genius

But I like to think I'm pretty smart, and analytical.

Anyway, of course the way the question is asked has a lot to do with the responses.

There's something called push-polling where the questions are designed to get a specific response. Dubya's campaign is a master of this, as they did it to Senator McCain in South Carolina in the 2000 Primaries (would it change your opinion of Senator McCain if you knew he was father to a bi-racial child?*)

But yes, I know a lot of people who don't like Bush, but aren't sold on Kerry yet either, so ask them if they think Bush should get a second term & they'll say no, but ask if they think Kerry should be president and they may say no to that too.

The yes or no question has no alternative. And sometimes doesn't account for uneducated people or people who may interpret the question differently.

on rnc.org there is a question
"Would you support a jobs and growth plan that gives taxpayers an average $1083 more a year"
Most people would want to say yes because they aren't paying attention to the word "average"

CNN.com often has poorly worded questions for their polls.
Example:
"Do you think the U.S. should share some responsibility for Nicholas Berg's death"
Yes or No. Well what if I think the U.S. is totally responsible (I don't, that's a little extreme, but hypothetically) - I object to the word "some" instead of "full." So, I vote "no." CNN.com has had far more flagrant examples.

Can you tell I used to agonize over True/False questions on tests?

* Senator John McCain and his wife Cindy adopted a girl from Bangladesh

[phantom power]

5. also poll questions hardly ever deal with realistic tradeoffs

For example: Would I support a jobs and growth plan that gives taxpayers an average of $1083 more a year?

Well, all things being equal, sure. But in real life, actions always have consequences. Does it increase the deficit? Does it take money away from something I think is important?

Poll questions hardly ever get into that kind of trade-off discussion, and tradeoffs are the whole point of policy.

[troublemaker]

6. You are correct that phrasing of questions steers answers

Perfect example in the recent CBS poll, showing that 54% say Iraq War was not "worth it" but 54% also say it was not "a mistake."

That said, the greatest power of polling is the ability to ask the same question at different times. If in a month Bush moves from 55% approval to 45% approval in the same poll that shows a real trend, even if nobody can agree what the heck it means to "approve" of the job someone is doing.