I've created this thread for those who like to create projections based on more than one poll (i.e. people who use weighted moving average, rolling averages or polls-of-polls).
I'm going to kick the discussion of with my projections based on Samplemiser
1. Hillary 2. Obama 3. Edwards 4. Biden 5. Richardson (raw polls)
1. Hillary. 2. Edwards 3. Obama 4. Biden 5. Richardson (raw polls adj for caucus rules).
http://thepoliticaltipster.wordpress.com/2007/12/22/why-hillary-should-be-smiling-for-now/
First, forget early national polls on the nomination.
Second, its extremely difficult/expensive to accurately survey for caucuses.
Third, all polls are not created equal. For example, Mason-Dixon is far better than ARG.
Fourth, polls are often mere "snapshots" which can change in a short time span.
So, in conclusion, while looking at several polls is useful, to equally weight ARG and Mason-Dixon (to use one example) is to skew the results. Its like adding dirty water (ARG) to clean water (Mason-Dixon).