quade 4 #1 August 10, 2004 Hey, they've put up a new web page that summarizes the national election polls pretty nicely. Even does the math for you on the margins (which is something they used to make you do for yourself). Anyway, give it a glance every now and again. http://www.pollingreport.com/wh2004.htmquade - The World's Most Boring Skydiver Quote Share this post Link to post Share on other sites
FallRate 0 #2 August 10, 2004 Polls are interesting, but I prefer to ignore them. I wish more politicians would do the same. That way we might get the straight scoop on what politicians (from all sides of the political machine) actually believe, instead of a laundry list of emotional keywords and phrases. Aid: "Sir, you must use the term "Family Values" no less than three times during this debate." Candidate: "Umm...ok...why?" Aid: "Our data shows that this phrase generates favorable response among our primary target audience for this debate. But be careful, the data also shows that if you use the phrase five or more times the average viewer's indulgance of the imagery this phrase invokes will wane." Candidate: "So...what exactly is it that I mean when I say 'Family Values'?" Aid: "It means approximately 3.6 percentage points." FallRate Quote Share this post Link to post Share on other sites
quade 4 #3 August 10, 2004 Well, it's true that the only -real- poll that counts happens the first Tuesday after the first Monday in November, but . . . that said it's "interesting" to watch the trends and reactions. I guess if you aren't really interested too much in the "game" then you can sit by the sidelines and choose not to watch. Then again, I can't fathom why anyone who wasn't into the "game" would venture into this Forum in general and this thread specifically.quade - The World's Most Boring Skydiver Quote Share this post Link to post Share on other sites
FallRate 0 #4 August 10, 2004 I find the game to be interesting. I just don't worry much about the halftime score. FallRate Quote Share this post Link to post Share on other sites
Balls 0 #5 August 10, 2004 Where does it show the Margin of Error? There is really no point in looking at polls when they are this close because the margin of error can swing it either way. Just wait till November.---------------------------------------- ....so that his place shall never be with those cold and timid souls who knew neither victory nor defeat." Quote Share this post Link to post Share on other sites
quade 4 #6 August 10, 2004 Margin of Error is shown on the details pages. The new page gives summaries for easy comparison, but the details are all still there if you want 'em.quade - The World's Most Boring Skydiver Quote Share this post Link to post Share on other sites
turtlespeed 212 #7 August 10, 2004 If that poll comes to be true . . . Life, as Americans know it, will take a slide into the nether reigons of hell.I'm not usually into the whole 3-way thing, but you got me a little excited with that. - Skymama BTR #1 / OTB^5 Official #2 / Hellfish #408 / VSCR #108/Tortuga/Orfun Quote Share this post Link to post Share on other sites
crwtom 0 #8 August 10, 2004 Quote Life, as Americans know it, will take a slide into the nether reigons of hell. This is priceless - right up there with those road side bill boards in the deep shouth that warn you to repent because hell is near and only Jesus can save you. Maybe we should issue a "purple doomsday alert" or something like that. Thanks for the splendid entertainment. As to polls' margins of error - if you look at it (a little naively) as a binomial distribution with N samplings and an overall probability p (think of it as true total percetages) between two outcomes then the variance of the sampled percetage is sqare root of p(1-p)/N. In the give example we have roughly p=0.5 and I guess the standard margin of error is twice the variance. That makes it appoximatiely the inverse square root of N - with N usually around 1000 in these polls you get indeed the 3% or so that you find in most of the polls. That also means that if you look at averages of, say, four polls (sized around 1000) you may think of that as one poll with N=4000. But four times the N means half times the error putting it at somwhere around 1.5%. It'd be cool to have some dynamic averaging algorithm feeded constantly by all of the current standard polls. It wouldn't require much to do that since the data is already there - just someone with a little time on their hands and/or a little bit of computer skills. You'd have a sort-of smoothened out polling product with a margin of error of possibly less than 1%. Only drawn back is that it'd just "run behind" the other polls by a couple of weeks since those are not conducted simultaneously. Cheers, T ******************************************************************* Fear causes hesitation, and hesitation will cause your worst fears to come true Quote Share this post Link to post Share on other sites