General Question

Blackberry's avatar

How accurate are polls?

Asked by Blackberry (31903points) May 19th, 2010

There is always some poll that people in the media use like “46% of the american people want this…”. But of course there can’t be such a widespread poll of everyone in the U.S. So how many people take these types of polls? Even something as widespread as the census or a presidential election isn’t that accurate because not everyone participates.

Observing members: 0 Composing members: 0

8 Answers

CaptainHarley's avatar

Depends upon who conducts them. I woudn’t trust any poll conducted by either of the major political parties, but Harris Polls and the others are relatively reliable and honest.

ragingloli's avatar

That depends on
– How many people are being polled? the smaller the sample size the less reliable the results
– By what criteria are people being selected for the poll? if you only poll conservatives/liberals, the results are pretty much fixed
– How are the questions worded? you can construct questions in such a way as to coax people into the answer that you want them to give
and last but not least:
– How are the results interpreted?

For example, take the polls that say that the majority of Americans disapprove of the healthcare reform bill as it was passed.
What they do not tell you is that this majority also includes the large number of people who think the bill does not go far enough, e.g. the people who wanted a public option/single payer system.

marinelife's avatar

Polling is a science. It is possible to get a reliable result for the whole country by strategic sampling and well-written questions.

It is possible to also use those techniques to present slanted results. So it depends on the reputation of the polling agency.

Blackberry's avatar

Hmmm, as I suspected lol….So I guess the question is: What polling agency is the most accurate and unbiased?

aprilsimnel's avatar

Well, as Mark Twain once said, there are “lies, damned lies and statistics”. Granted, he misattributed the phrase to Disraeli, but whatever.

Questions can be manipulated or leading and numbers can be manipulated of faked to reflect the biases of the pollsters, which happens all the time, never mind the margin of error. The only way you’d get anything down accurately is if everyone answered non-biased questions honestly, none of which happens, so selection bias and response bias are big problems! Some people don’t have phones, so there’s coverage bias.

I used to take health surveys for three different states through the UW-Madison Survey Research Lab, and my job was the refusal converter. Ugh. In other words, after a surveyer called a qualifying household and was hung up on (or cursed out or otherwise refused), my job was to call that household back to convince someone to do the study. The more answers we got, the more accuracy, supposedly. My conversion rate was ~32%. I hated every second of that job, but it paid stellar wages.

The Pew Research Center is a watchdog think tank for pollsters. You could start there.

bunnygrl's avatar

@aprilsimnel first thing that jumped into my head was that lies, damned lies quote lol and also, I just squeeeeled at your icon completely adorable :-)

I don’t usually put a whole lot of stock into polls and percentages, since if you read the very very very small print in ads you’ll find that the “92% of those asked said this shampoo left their hair fuller/glossier/sleeker etc” is a bit less impressive when they say their test group was 120 people. That having been said, sad politics junkie that I am, I recently stayed up all night to watch the general election results announced as they happened. At the beginning of the long long night, they announced an exit poll and all of the commentators in the studio basically poo poo ed it as nonsense, well it was almost spot on. The lib dems did lose seats rather than gain, and the conservatives ended up almost exactly where the exit poll put them. I was waaaaay impressed I have to say. So, as others have said, I suppose it depends on the number polled, who’s asking the questions, and which questions are asked.
hugs xx

perspicacious's avatar

There is always a margin of error stated with a poll. The real answer goes to how questions are asked. Sociologists are very good at getting the answers they want by the way they frame a question.

mattbrowne's avatar

Depends on the quality of statisticians. There are a lot of amateurs out there. Reliable polls cost a lot of money.

Answer this question

Login

or

Join

to answer.

This question is in the General Section. Responses must be helpful and on-topic.

Your answer will be saved while you login or join.

Have a question? Ask Fluther!

What do you know more about?
or
Knowledge Networking @ Fluther