Survivor Polling Data
There's a pretty good breakdown of the info at the Gallup Organization's homepage http://www.gallup.com/help/FAQs/poll1.asp
but the quick version works like this:
1. Work out a series of carefully-worded, non-leading opinion questions with multiple choice answers. Usually the answers are opposites (yes/no, more likely/less likely, approve/disapprove, Bush/Kerry, etc.), but you leave a third option open for undecided/I don't know answers. Ask however many questions you want, but more questions gives you better data to work with (to a point, anyway). (We should, all of us, probably kick in a few poll questions to help you out there, but I'd bet you can pull some from a few polls out there or get some forum folks somewhere to give you a bunch.)
2. Choose a target demographic ("Americans of voting age," "UW Students," etc.) and select a random sample of them to call. You typically poll 1,000 people (well, until you get 1,000 answers) - it's accurate to +/- 3% and a larger number doesn't improve that accuracy much, but does cost more.
3. You hire people to make calls and conduct the poll. These people ask the questions and record, anonymously, the responses.
4. When you successfully poll 1,000 people (the most cost-effective number; it has an accuracy of +/- 3%), you put your data in a report with colored bar graphs and percentage numbers ("60% in favor, 30% opposed, with 10% undecided"), as well as an analysis of what those numbers mean ("Steve, people totally hate you").
WHO is doing this polling??
Anyone. Seriously, there are a few companies that exist to do polling (Gallup is the most well-known), as well as private pollsters (Joey Lucas on The West Wing, for example). Every news agency does their own polls, too, and most major corporations have their marketing firms do polls for them.
We'll probably see CNN, MSNBC, etc. polls on the Survivors, but we'd want to do our own, too. That's so we can compare numbers and also because each poll will have a few unique questions. We could conduct the poll ourselves ("internal polling") or hire someone to do it for us; the later is considered more accurate, so it's what I'd lean for, esp. with the Foundation just starting up.
