Just curious..
Do not have any special access to or knowledge of how each of these polls is conducted. Personal observation leads me to believe; every poll uses its own arcane technique loosely based on science, most polls are crafted to return desired results (push polls), most polling questions are purposely ambiguous, and the interpretations of poll question responses are infinitely malleable.
The only polls which I would trust are the ones which; specify up-front solid, unambiguous assumptions/conditions of the poll, only ask unambiguous questions, ask a complete range of questions based on the assumptions/conditions of the poll, provide a complete range of unambiguous responses to each question, poll a significant sample size (say 10,000+ for a Presidential election), spread that sample size over a population appropriate to the poll (i.e. for Presidential elections spread over all 50 states in proportion to their Electoral College votes and in a truly random fashion within each state), present all of the above at the conclusion of the poll, and engage in a minimum of analysis of the results.
In short, I trust no poll because none meet my conditions of a valid poll.
I believe the best we can derive from today's polls are very short term indicators. Other than that, as a good poker player plays the other players, we have to play the pollsters and be aware of their "tells".
Wish I had a better answer for you. Hope some Freeper statistician can pick up on your question.