DescentBB

Just...wow.



http://www.zerohedge.com/news/2016-10-2 ... versamples

Someone in this thread doesn't know enough about statistics to understand what is being said here.

how is this playing dirty? Sort of seems to be an incomplete item, looks to be a cover page with a deeper analysis buried behind it. Routine stuff. I got these every other day, from Aug 15 to election day in 2008, and I wasn't far up the food chain. Still I got data for targeting purposes, for me to dovetail with my outreach data for my district. That isn't cheating, it's called running a damned campaign. What am I missing here, NS?

the zerohedge article babbles about sample distribution bias. All polls have one. They are designed based upon turnout models. Folks way more expert than any of us analyze various viewpoints and design the targeted respondants sorted into various groupings, which can be amazingly detailed demographicaly. Anyhow, to overgeneralize for you: There are roughly 36% registered Dems, 25% registered Reps, a little more than 30% independent or third party. Generic, typical turnout models tend to have about 40% Dems, 30% GOP, 30% indies, because the registered party loyals tend to turn out a bit more than unaffiliateds. This year, a glance at the major national polls seems to show two distinct turnout models in use. One seems to assume a very motivated Trump turnout, and those are the polls you see than the fine print shows to be sorted roughly 40%D, 40%R, 20%Ind(assume general disdain in the latter?), the other clump seems to run around 45%D,30%R,25%I. The first group are those polls you see that, at the moment give you Trump within margin of error, maybe outside it by one or two. The latter are the ones showing Hillary whomping Trump by 8,12, 14%. We won't know who's right until election day, and the wide spread of predicted turnout models is sort of wild, but pretty much everything about this one has been different. You can bet both camps have(or ought to have, I suspect Trump may not)extremely accurate state by state polling data, that is an order of magnitude better than the public stuff. They pay for it, usually with multiple polls pooled and anaylzed. How is all this playing dirty? Beats me.

Its in an email from democrats, so it must be evil-mail instead of electronic-mail.

Just because you can't understand the terms these democrats are using does not automatically mean they are doing something bad. That is a major problem with the conservative movement these days, they attack smart people because they say smart people get away with evil things. Well I've got some news for you: being that stupid is the real evil, especially when you do it on purpose like that.