These robo-tracks are pretty useless IMHO.
The "zero sum" effect is a bit strange.
We know that all his June state robo-tracks add up to a dead heat,as they are the parts that make up the National track whole.
The "changes" from state to state are random noise.
In reality you will almost never gain 4% in state X, while at the same time losing 4% in State Y.
If a candidate gains 5%, the states all move +/- 5%. It might be 3% or it might be 7%, but the "tide" goes in the same direction regardless of where you are.
There are ALWAYS exceptions (WV in 2000 for example) but the tide goes the same way 99% of the time.
The "Robo Track" state polls are random Noise IMHO.
True, but their could be other examples. For instance (and I'm just using this as an example so its not real) lets say in May Kerry stumped a ton in Ohio and Pennsylvania and Bush stumpped a ton in Florida & Michagin, but in June Kerry stumped heavily in Florida, Bush stumped heavily in Ohio. In May Ohio & PA could have an unrealisic high for Kerry and Florida & Michh could have an unrealistic high for Bush because if just rollover effect from the stumping in those states. PA & Mich may see a change back toward the more normal status (PA in Bush's favor, MI in Kerry's in June because the May stumpinf factor no longer impacts those states. But Florida could turn to an unrealistic high for Kerry & Ohio could turn to an unrealistic high for Bush in June because the heavy stumping factor.
Lets say the national polls from May to June have very little change, most states may see a small change in either direction within the MOE or statistical range +2 or -2, but others such as these 4 could see rather large changes, especially 2 of them due to the impact a candidate has when they stump heavily in a state during any given period.
