Who are they polling?

The media receives this sort of information from commercial think tanks that conduct public opinion surveys on salient political issues or recent news-worthy events. “Gallup” should ring a bell.

Edit: the selection of who such companies and organizations poll is grounded in a scientific principal concerning their sampling universe—in theory, the sample population must be selected as randomly as possible.

Yeah – polls are stupid – the Iowa Electronic Markets, however, rate the likelihood of specific candidacy success because participants, in essence, buy and sell shares based on the likelihood of a specific outcome – with ridiculous accuracy as to the actual results.

Polls are often oddly inaccurate because they don’t really require anything be put at risk…it’s a question in a vacuum, without discussion. And responders, as @Batsh1t_Cat_Lady, are often self-selecting which decreases the randomness of the sample set.

Scientifically conducted polls of a large sample group are generally accurate unless they try to measure things people are ashamed to admit. Polling gets a bad name though because numerous partisan groups conduct polls that are designed to get a specific answer. They have no intention of reflecting actual public attitudes.

For instance, suppose before the big healthcare vote you wanted to poll the opinion of average Americans. If you contacted a randomly selected group and asked them if they supported a change in law so that the government could decide when you no longer get healthcare, you would get an overwhelming no. If you asked if you wanted to prevent healthcare insurers from canceling your policy when they learn you are seriously ill, you’d get an overwhelming yes.

A real world example actually happened on healthcare. The poll asked 3 questions.
When thinking about the current proposal for healthcare reform:
1—Are you opposed to the proposed bill?
2—Do you think the proposed bill is about right?
3—Do you think the proposed bill doesn’t go far enough?

They reported that 67% of Americans opposed healthcare reform, because they lumped together those who didn’t want it at all and those who wanted even more change. In fact, the majority either thought it was good as is or wanted more change. But you had to dig deep into the study to find out what the people responding really answered. The spin masters used the results to further their own ends.

Who do they poll? They select a random sample from the population they are interested in finding out about. A representative sample for a nation the size of the US is a little over one thousand people. They usually select phone numbers at random, and then call and call until someone answers their questions. If too many people refuse to answer, the poll results are likely to be skewed.

You have to remember that they are polling people that represent millions of others. Your friends may all disagree with the poll, but the poll can still be right, Your poll of your friends is not a random sample. It’s called a “convenience” sample. Thus, you can not generalize from your poll and have anyone believe you are talking about anyone other than your friends.

I have been polled several times. The last time by an outfit in Canada that seemed to be looking at attitudes about politicians and the economy. I have no idea who their client was. It was not a push poll (where a candidate runs a “poll” that is really designed to “educate” people about the candidate and thereby encourage them to support the candidate). It seemed neutral.

Being liberal, I love to answer polls, because they do influence people. If my opinion is repeated by a few other people in the poll, then it may appear that the electorate is more liberal, which may encourage others to be more liberal. But, if the sample is properly selected, I won’t have that effect. Still, there’s always a chance.

@wundayatta Just to be clear, they weren’t friends. They are people at my job from all walks of life. It was a pretty random sample, but I see what you’re saying.

Thanks everyone for you answers. I had no idea.

I agree with @ETpro.
Polling also does one other thing. They tend to send pollers to conservative places. So the not so conservative are rarely asked their opinion. Also the conservatives are much more likely to actually answer polls. Unless a tv crew is around. Then every one wants to be a star. LOL

@Facade Technically, that does not count as a random sample to a statistician. Everyone works at the same place. There could easily be something about working there that biases their thinking. Group think is one thing. The impact of a policy on their employment is another. These are not random people at all.