Summary of polls at RCP: Kerry wins

[pabloseb]

This is the average of polls in key battleground states from realclearpolitics.com (right wing site but quite complete in terms of polls):

• Florida: Bush + 2.2
• Ohio: Kerry +2.0
• Pennsylvania: Kerry +2.4
• Wisconsin: Kerry +0.3
• Iowa: Bush + 1.3
• Minnesota: Kerry +1.7
• Michigan: Kerry +3.2
• New Mexico: Bush + 3.3
• New Hampshire: Kerry +2.7

Assuming that all the other states have the same result as in 2000, Kerry would win the EC with 272 electoral votes. And this is before taking into account the undecideds, the huge turnout especially among the young and minorities, etc.

:bounce: :bounce:

[pabloseb]

1. Kick

:hi:

[polmaven]

2. We're not going to start believing

realclearpolitics just because they are currently in our favor, are we?

They were just a tad wrong in 2000 when they said Bush would have 446 EV to Gore's 92.......Ummmm........We need a little balance here.

[pabloseb]

4. I'm just quoting their poll averages

They do take into account more polls that any other site. The fact that the editors are right-wing is not relevant.

Their predictions on who will win are a whole another matter. Although this year they're not predicting any kind of landslide.

[polmaven]

5. Just don't get too excited about it.

This particular site is not a good one to count on.

[endnote]

3. Polls are BS. Polls WITHOUT MOE information are plain stupid.

[jefferson_dem]

6. I so want to feel good about this but recall RCP's 2000 projection

Their track record is...less than perfect.

http://web.archive.org/web/20001110051000/www.realclearpolitics.com/Polls/polls-EC_ALWAYS.html

[shockingelk]

9. if they're off my as much

and in the same direction, it will be a Kerry landslide.

[foo_bar]

10. they were half right

The media's fantasy of Bush winning the popular vote and losing the electoral college is not going happen.

[sundancekid]

7. trouble is that averaging usually assures that a measure goes to the MOE,

and your table does exactly that ... the MOE in all these states is around +/- 4 ... therefore, NOTHING can be concluded in any credible sense

[DemocratSinceBirth]

8. au contraire...

the more polls you have the less your margin of error...



If you have four polls asking the same question of 1,000 people each your moe is about 1%


if you have one poll your moe is about 3.9%...


It's late or I'd give a fuller explanation...

[sundancekid]

11. I appreciate your point (see formula below) BUT it assumes the same method

for selecting the sample (i.e. registered v. likely voters, %mix of R's, D's etc..) and the various polls used to average do not have identical criteria; my point was metaphoric rather than mathematical ... and when one averages apples and oranges and bananas, what you get is?????

Note: usual formula is 1/(.04)squared = a sample size of 625 for a +/- 4% MOE; therefore, for approximately a 4000 person sample size,
working the formula backwards it would = 1/(.0158)squared, thus producing a 1.58% MOE -- and that decrease assumes identical methodology used on all 4000 persons in the sample, which we do not have here.