[gssixgun]

Why does the main stream media ignore this poll, including AOL headline news itself... http://news.aol.com/political-machin...ll-sept-11-18/
This poll will have over 300,000 votes by the end of the week, every week since it started it has never been closer than 60/40 (McCain) and the lowest number of votes it has had, was 180,000, that I have seen in 3 months.....
But we get polls reported on National news that have 500-5000 people and this is what is being reported on.....
I am only asking WHY??????
BTW the only reason the number this week is so low is because AOL pulled this poll for over a week and it just came back up today....
You might say because there is no proof that these are registered voters????
I could accept that, but really can this be all kids voting for McCain???? (that dog just doesn't hunt for me)
You can't vote twice BTW from the same computer, try it yerself....

What say you????

Also let's keep this on course, and not take it into a my dog is bigger than your dog discussion that goes nowhere.....

[hoglahoo]

because it's an online poll I guess

It reminds me of all those straw polls Ron Paul won by huge margins before the primaries earlier this year

[JMS]

It apparently doesn't suit the purpose of the media!

[igitur55]

How reliable is it? McCain winning by a large margin in DC makes me doubtful ... .

[BFMC AJ]

First off let me say I am a McCain supporter.

Having said that, this poll has no statistical validity at all. It is not a random sampling, and is easilly swayed by mass e-mails asking supporters to vote a certain way.

Having said THAT...I don't trust polls in general because it is far too easy for the poll designer to influence the results with the questions asked or even the verbage or tone used in the interview.

[Tennessee Dave]

Originally Posted by JMS

It apparently doesn't suit the purpose of the media!

I think this is right on,

[NeitherMightier]

Voluntary-response polls are meaningless. There's no mechanism to even TRY to ensure that the respondents are a representative sample. Also, CAPTCHAs, such as the one before the vote, which _ostensibly_ filter out automated responses to web apps like this, have grown unreliable over the past two years or so.

Simply put, it's not a scientific poll.

Alexander

[Nord Jim]

Lee and others are correct.

Internet polls are what statisticians refer to as "self-selecting." That means that the sampling isn't at all random. People choose to answer, and this is affected by who visits the site and if they have an axe to grind. It's always easier to get negative feedback than positive, for example, in a self-selecting poll.

You could ask the exact same question on Foxnews.com and CNN.com and get completely different results. Such polls are useless. Web sites actually run these polls as market research, not for political information.

To reply to the original musing about polls that have 500 or 1000 responses: It has to do with how the margin of error is calculated. MOE is determined by an equation that is the number 1 divided by the square root of the size of the sampling. If you do the math, you find that a sampling of around 625 responses gives you a result that has a margin of error of plus or minus 4%. To get a margin of 3%, you need a sampling of about 1108. That's a popular sampling size because to get an even lower MOE requires so many more responses that the benefit isn't considered worth the effort and cost.

The bugaboos of polling are the size of the sampling, whether the sampling is truly random (which can be difficult), and if the question is asked in a neutral, clear manner.

j

[JMS]

I have to say, I don't think the media is concerned with accuracy!