Reply #14: You weight demograghic variables...not *sample* size. [View All]

[Petrodollar Warfare]

While exit polls do conduct "weighting" for various demographic variables (ie. race, gender, etc), you can not weight polls for "people." Afterall, 1 person polled = 1 person polled regarding overall sample size. If N = 1963 voters/people at 12 midnight, and N = 2020 voters/people at 1am, how does one ascribe different "weights" to the two sample sizes to show the overall variance/shift in total Bush vs. Kerry percentages? Answer: You can't - at least not using the laws of mathematics...perhaps its fuzzy math? BTW, the 2000 election has nothing to do with the particular results that I posted.

Remmember, exit polling methodology is a very mature science that has been conducted accurately around the world for the past 3 decades, and even in this country throughout the 1980s-1990s, at least up until 2000 election in Florida, and now it appears "off" on multiple states - skewing in one and only direction. The variable that has changed is not exit polling metholodogy, but voting technology. A someone who has studied statistics (in MBA school) and Information Technology/INFOSEC during a 2nd Master's degree, I find this type of analysis...interesting...but not surprising. Politics has always been a high-stakes game.

Anyhow, my mathametical analysis is based purely whole integers that are not is any way weighted during exit polls. (N = sample size, which can not be weighted in the overall analysis) I hope this info helps.