Another thought: how difficult is it to predict what a "likely" voter is for this 2016 contest? I believe that there are factors in play that are likely to render the existing categorization of a "likely" voter obsolete.
This is ALWAYS the x factor, cause they never know. No one in 2008 thought turnout would be what it was. McCain was close for a while % wise, but with greater turnout, the percentages got to be huge margins.
It’s the reverse here. The primary suggests that GOP will have a much higher % of “likely” voters so the D/R splits on all these polls are useless. If it’s only 4% (vs. the 6-10 they’ve been running-—or even higher), then Trump wins huge.