You want polls, we've got 'em
- Thread starter jlabsher
- Start date
One poll has Bush up by 12 points, another has him and Kerry neck and neck. That's why I don't really put a whole lot of stock in polls - different organizations show different results so, to me, they're worthless.
B
Bruzilla
Guest
Thanks for the confirmations that Bush will win, but I thought the best part of the site was the link for the "W" ketchup. "You Don't Support Democrats - Why Should Your Ketchup?" That's funny!
http://www.wketchup.com/
http://www.wketchup.com/
about as dumb as FREEDOM fries
I do love enterprising people who see a niche and fill it. Keeps them off the breadline and we get things like Pet Rocks and W Ketchup.J.South said:
DID YOU KNOW that, at one time, the internet was a "fad" that "wouldn't last"? Mighty oaks from little acorns grow.
SamSpade
Well-Known Member
vraiblonde said:
As I mentioned before, and as I am sure you know - the difference in polls is based on the assumptions behind the sample used.
Most of them - most of the better ones - do NOT just get a random sample of the country. For one thing, completely random would get lots of ineligible voters, unregistered voters and so on. It's more intelligent to use "voters" as your universe to draw your sample from.
Many also use "likely voters" - people who intend to cast their vote, people who have voted in the past. Someone who votes for his Congressman in an off-year election will probably vote for President.
THEN comes the part where pollsters begin to diverge - should they randomly sample likely voters? Or should they weight their samples according to the disribution of Republicans, Democrats and independents? What IS the current composition of likely voters?
For example, let's say your state is made up of 38% Republicans, 42% Democrats and 20% everything else. Assuming these are *likely* voters, that's what the crowd headed to the polls on Election Day will resemble. So you create your sample based on that.
Now I've heard some conservatives - and I'm a conservative, but one whose job is statistics - some conservative NITWITS say "why don't you just poll everyone? If you get more Republicans, it just means more people registered Republican since the last election!". Yes. Wishful thinking. NEVER happens. Yeah, you get a lot of new registrants, but never enough to really change the demographics of a large region all that much. It's not just bad social science, it's bad math. You don't base your sample on your sample.
So the difference in the polls are, the different assumptions made by the pollsters, and a lot of them are wrong. If ANY pollster shows huge swings in the course of a week or a few days, DON'T believe them. They cannot be accurate. By this time in an election, each side has its base firmly in hand. Unless something catastrophic occurs, there won't be big changes in the numbers.
What worries me is, most undecideds usually favor the non-incumbent. And they usually make up their mind in the last couple days if not on the day of the election. A CLOSE race will favor Kerry, especially if his supporters are more urgent about getting new voters to the polls. I've never known Republicans to be the slightest bit successful in drumming up crowds of new voters. They don't "get out the vote". Secondly, the overseas vote this year will likely go massively against George Bush, because overseas voters are mostly non-military, and they live in nations hostile to Bush, and their opinions will reflect those hostilities. It will be close, but unless Bush opens a big lead, he'll lose. He needs a big slam dunk in the next two months, or he can't pull it off.
D
dems4me
Guest
Vince said:I'm sticking to my original prediction. Bush by 75%And you should all mark this as a quote because when it comes true I want to see everyone doing this
Will
suffice 