Kasz216 said:
Yeah, in it's state samples for congressional races. Presidential election samples and state election samples are totally different formulas. Since we're on VGchartz... a good example is well... vgChartz. They have a different formula for each console, because the breakdown of sales isn't consistent. Toys R US sells a higher percentage of total wii's then it does PS3, Amazon sells a higher percentage of Sony based stuff. etc. Most likely Rasmussen's sampling group is great for simulating the federal but sucks on the local level. They probably have a great representative sample of the country that isn't geographically located like you would expect within the country.
Rasmussen also had the most accurate presidential polls in 2004. http://www.slate.com/articles/health_and_science/human_nature/2004/12/lets_go_to_the_audiotape.html
|
Even better because we have a electoral college. Which means national polls are nothing but corporate news hype train.
