kane

When I say science I mean the methodology that many polling companies use. There are some that have a formula of how they pick the people the poll, what questions they ask and how they ask them that allow them to be pretty accurate. Others simply dial 1000 random numbers and are not as accurate.

Polls are, historically speaking, pretty accurate. If you look back over the past presidential elections and you look at the state by state polls they were pretty damn close to how that state ended up voting. The same can be said for most of the national polls. Of course they are not exact, something that is subject to people's changing minds is never going to be exact, but many of them can be pretty accurate.

Exit polling, until the last presidential election, has been extremely reliable and even then it was pretty spot on with the exception of a few places. The UN actually uses polling, especially exit polling, to determine if elections held in third world countries are corrupt or not and in many cases they use the same the companies that provide the presidential polls. Here is an article on how exit polls are often used to make sure the vote count is accurate http://www.nytimes.com/2004/10/17/we...ew/17plis.html

Here is a good site that shows the recent presidential elections and how the different polling companies did as far as their accuracy.
http://stolenthunder.blogspot.com/20...-national.html

With the vast majority of the companies listed they were within their margin of error on accuracy.

Here is gallup's numbers on the past presidential elections. http://www.gallup.com/poll/9442/Elec...Elections.aspx again, they are almost always within their margin of error.

Polling is not an exact science, but it is a science in that it has a system that it uses and a formula that it follows.