I don't put much faith in polls but just in case you were wondering...

 

ALL POLLS ARE FROM THE LAST WEEK

 

FLORIDA:

Research 2000---- Kerry by 1%

Miami Herald------ TIED

Rassmussen------- TIED

CNN/USA Today--- Bush by 9% (???)

Insider Advantage- TIED

Survey USA------ Kerry by 2%

American Research Kerry by 3%

 

OHIO:

ABC News----- Kerry by 3%

SurveyUSA---- Kerry by 2%

Fox News----- Bush by 2%

CNN/USA Today-- Kerry by 6%

Ohio University--- Kerry by 6%

American Research Kerry by 2%

 

PENNSYLVANIA:

Quinnipac Univ Kerry by 6%

Muhlendberg Kerry by 2%

Rassmussen Kerry by 3%

Keystone Poll Kerry by 5%

American Research Kerry by 3%

 

Could Kerry get a clean sweep of all 3??? Florida is obviously the toughest one.

http://www.electoral-vote.com/

 

From what this site says Bush can lose PA and Ohio and still win if he picks up Arkansas.

no doubt about it.

 

Kerry would need to pick up Ohio, New Hampshire and protect either Iowa or Wisconsin in order to win.

 

Personally I don't think Arkansas is even up in the air. I think that's a 5 pt Bush win.

Yeah, due to the margin of error, pretty much all those polls say is that there is a tie in all three of those states. I don't know about you guys, but I'm definitely in the group that doesn't think we'll know the winner Wednesday morning. There's so many close races in states that you know the chances are high there will be a problem somewhere. You know what else makes me completely ignore polls. Did you know to get a margin of error of +/- 3% only about 1000 people need to be interviewed? That's crazy.

Saying a three percent margin of error. It usually means that there's a 90 to 95 percent certainty of a 3 or 4 point margin of error. That means that a couple times out of twenty, you're likely to get bad data.

Well at my high school, Kerry defeated Bush 238 to 118.....

 

I'm in Iowa, BTW....our high school is good trend data.....

Saying a three percent margin of error. It usually means that there's a 90 to 95 percent certainty of a 3 or 4 point margin of error. That means that a couple times out of twenty, you're likely to get bad data.

Yeah, I know. It's called a 95% confidence level which means that the researcher is 95% confident that the actual results will be within the margin of error. But even with a 95% confidence level, when the candidates are within 2-3 points of each other the poll pretty much tells you nothing. It could say Kerry has 49% and Bush has 46% when in fact it's the exaxt opposite and Kerry has 46% and Bush has 49%. Florida in 2000 is the perfect example of this.

 

Furthermore, these polls are usually being done by telephone interviews. About 97% of the country has telephones, but you have to think are there differences between the people who agree to the survey and those who don't? What time of day is the interview being done? How many of the people doing the survey are actually going to go out and vote?

Ahhh, good, I was just coming here to lambast the polls and how, with the margin of error, they mean next to nothing. Glad to see we have some other stats people keeping up the pace.

Furthermore, these polls are usually being done by telephone interviews.  About 97% of the country has telephones, but you have to think are there differences between the people who agree to the survey and those who don't?  What time of day is the interview being done?  How many of the people doing the survey are actually going to go out and vote?

The telephone issue is even more problematic than in the past. An NPR piece last week noted that cell phones are largely ignored by the pollsters because tthere is not a readily accessible centralized database of cell phone listings. There is a growing number of people - mostly young profesionals - who have entirely done away with their land lines and are just using cells. So any traditional polling practice that misses these segments is apt to give a skewed view of public opinion.

The telephone issue is even more problematic than in the past.  An NPR piece last week noted that cell phones are largely ignored by the pollsters because tthere is not a readily accessible centralized database of cell phone listings.  There is a growing number of people - mostly young profesionals - who have entirely done away with their land lines and are just using cells.  So any traditional polling practice that misses these segments is apt to give a skewed view of public opinion.

I agree completely, I was watching the news over the weekend and they had some pollsters on defending the samples they collected for the polls. And the lady said she had no reason to believe that just using landlines would miss out on people that were substantially different from the general public. Well, obviously, I think it's a flaw if you're missing an entire group of people. All of my friends from college and here in grad school only have cell phones (many in the swing state MN)--and I would argue we're a slightly different breed.

 

Ah, the joys of getting a sample that is representative of your population. So, between sampling problems and the stats stuff I would say that polls are good enough to use and kleenex, but not much more than that.

Edited October 27, 200421 yr by ChiSoxyGirl

And the lady said she had no reason to believe that just using landmines would miss out on people that were substantially different from the general public.

Landmines?? Are we using those now?? Funny, I would think blowing up people is not the best way to get their information.

Landmines??  Are we using those now??  Funny, I would think blowing up people is not the best way to get their information.

Lol, sorry, no idea where that came from.

Ahhh, good, I was just coming here to lambast the polls and how, with the margin of error, they mean next to nothing. Glad to see we have some other stats people keeping up the pace.

cough*nerds*cough

 

And yes I followed every bit of the statistical analysis too

cough*nerds*cough

 

And yes I followed every bit of the statistical analysis too

I was forced to be a nerd! Stupid Political Science course requirements

I feel for anyone who has to take stats. I'm in the 2nd version of stats right now and god do I hate it.

 

But ya, I call it a CI too.