Actually, Zogby's tracking poll today (with Nader) shows Bush with a 4-point lead!

http://realclearpolitics.com/bush_vs_kerry_hth.html

Sorry, wrong link. Try this one.

Dammit, I'll get it right yet! Ace, feel free to edit these.

There!!
http://www.zogby.com/news/ReadNews886.html

I agree that 48-46 is 2%, but all the Zogby numbers I've seen for 10/14 show 48-44 Bush-Kerry. That'd be, um, let's see, yup +4%. Hope it gets bigger, too!

Ace, this is random and has nothing to do with this post, but just curious if you heard they're making Miami Vice into a movie. Michael Mann writing/directing and Jamie Foxx as Tubbs... couldn't find any casting info for Crockett. Just a heads up if you haven't heard.

Yes, actually, several posters pointed that out.

While you are in the update mode, note that 4 is OUTSIDE the MoE (2.9).

Yeah, I know you'll grumble a little at all the work, but inside, you'll be singing.

The most recent thought is that Zogby (the liberal pollster) is cooking the books in order to make it look like bush is ahead, if only to draw out democrats to vote in larger numbers.

//removes tinfoil hat

Not sure I buy it - this guy has no vested interest in either candidate, and his whole reputation could be shot to shit with a CBS-like move like that. However, all it takes is one pro-bush insider to catch whiff of something like that an *poof* there goes a credibility bubble.

Food for thought.

Yeah, I know you'll grumble a little at all the work, but inside, you'll be singing.

But the MoE is for each candidate. To find the MoE for the spread, double the MoE.

Actually, the MOE for the difference between two samples in NOT double the MOE of the individual samples. I don't have the formula on me, but a rule of thumb (hat tip, American Statistical Society)is to multiply the MOE of one sample by 1.7. Hence, the MOE for the delta between the two samples is approximately 4.9%.

I found a good primer by the ASA at:

http://www.amstat.org/sections/srms/brochures/margin.pdf

distributed intelligence - a beautiful thing

Don't forget the weekend effect in 3-day tracking polls. Every Sunday, Monday and Tuesday poll will reflect the over-polling of Dems on weekends.

Wednesday, Thursday and Friday will always reflect more honest polling.

Project for people: Go review the swings in Rasmussen. The weekend effect is obvious.

And Ace, "several posters" have mentioned it. Can a brotha get a shout out?

The above last question was a joke, in case it wasn't obvious.

Thanks for all the previous shout outs.

Why is everyone so negative on POE ?? Why couldnt Bush's lead be 8 pts.. Hmmm??? or split the difference and say 6 pts. Why should we assume all errors will benefit Kerry?? Just a thought.

What we are doing when we look at margin of error is saying that there is a 95% chance the true difference is the measured difference +/- MOE. The 95% is the convention that scientists use, but it is essentially arbitrary. In ballpark terms, there is probably a 90% chance that the Zogby poll measures a true lead for Bush over Kerry. For non-scientists, them's pretty good odds.

The big problem with the polls, of course, is not MOE but methodology. Does their sample represent a true random sample of teh electorate that will vote on election day? If it doesn't then the MOE doesn't really mean a damn thing.

poker me up