Internal polls should be better because they start with the voter data and they have their own metriics ofrom their GOTV operations, but they can get things wrong if they make bad assumptions. Romney's team assumed they would get their voters out and Obama wouldn't. They were wrong. It would be curious to see what assumptions the Trump team are using to show leads in NV and OH. I would bet that Clinton internals have her up in both.
It's like that recent upshot experiment where they gave the same raw data to four different pollsters, and each had a different result. Some showing trump up some Clinton up.
Internal polls may be better for people working in the campaign who get to see all of them, but obviously we don't get to see all of them.
From the Democratic side, we are selectively leaked the ones that look good (e.g. statistical fluctuations in their direction). From the Republican side, we get more because that ship is leaking like crazy...but we generally expect Democratic polls to be of better quality, because all the young people with brains are working for the Democrats.
But yes, obviously it's not good for Trump that he's down in 4 states that he needs to win, and his only possible path to victory is the FL/NC/NH/ME-2 eye of the needle to 270 (and hope there are no faithless electors). Also doesn't help that those states aren't the best correlated, so pulling the inside straight here isn't likely at all.
