One important aspect of almost all of the polls is that the people polled must fit the definition of "likely voters". Limiting it to them seems, on the surface, to be wise: if someone is unlikely to vote, then their choice means absolutely nothing.
But the details of the likely voters definition are important. Specifically, anyone who would be a first time voter is almost always NOT considered to be a likely voter.
The issues of abortion, LGBTQ rights, environment, and (in many states) referendums to relax marijuana laws have together have caused a spike in people registering to vote for the first time in many states.
Will they vote? History says no. But there would appear to be an unusual degree of interest in the under-30 crowd, reflected at least in voter registration numbers. If they show up then it is a potentially large voting block that virtually all pollsters are purposefully ignoring under the expectation that few of them will bother to cast votes.
Whereas most polls in the past have indeed underestimated Trump's support, this factor may be doing the opposite, as the under-30 demographic is strongly anti-Trump
But the details of the likely voters definition are important. Specifically, anyone who would be a first time voter is almost always NOT considered to be a likely voter.
The issues of abortion, LGBTQ rights, environment, and (in many states) referendums to relax marijuana laws have together have caused a spike in people registering to vote for the first time in many states.
Will they vote? History says no. But there would appear to be an unusual degree of interest in the under-30 crowd, reflected at least in voter registration numbers. If they show up then it is a potentially large voting block that virtually all pollsters are purposefully ignoring under the expectation that few of them will bother to cast votes.
Whereas most polls in the past have indeed underestimated Trump's support, this factor may be doing the opposite, as the under-30 demographic is strongly anti-Trump