.. is that it treats the poll positions in different states as if they were independent of results in other states.

The reason I emphasised the early states over the national standing is because they are, as you suggest, crucial to building momentum and thus influencing future states. Clinton's 18% national lead (just increased to 19%) is so much "paper profits" until she actually starts cashing it in in real votes in real primaries.

Real Clear Politics has Clinton's lead in Iowa now increasing to 3.3% (from Edwards) and 3.5% from Obama which is obviously still surmountable by either, but it is hard to see Edwards make up a 15% point gap in New Hampshire even if he does win Iowa. Edwards and Obama seem to be cannibalizing each other (when one goes down the other goes up) and so one of them needs to pull well clear of the other to have a chance of beating Clinton.

Of course Clinton could easily still lose both Iowa and New Hampshire but she is the only candidate who would remain viable after doing so.  Her 20% lead in all the other early states (bar South Carolina) gives her a huge cushion against possible "bandwagon" and other defections.

By contrast one of Obama or Edwards really has to pull well ahead of the other to sustain a viable challenge as they seem to be fishing in the same pool. So long as they are tightly bunched, Clinton wins.

Index of Frank's Diaries