Iowa and Nevada are closed caucuses and Democrat numbers are not recorded precisely, so let's consider the increases in Republican turnout in each of these states from 2008 to 2016:
The total US population grew 6% over that same period of time.
Even more suggestive of a Trumpian November triumph are the non-closed contests in New Hampshire and South Carolina, where unaffiliated voters may participate in either party's primary (but not both of them). The following table shows total votes cast in each primary and the percentages of each year's voters who participated in the Democrat and Republican primaries (D%-R%):
|State||Dem '08||GOP '08||Dem-Rep||Dem '16||GOP '16||Dem-Rep|
The contrast in South Carolina between '08 and '16 is astounding. In '08, more people voted in the Democrat primary than in the Republican one. Eight years later, there were more than twice as many participants on the Republican side as there were on the Democrat side.
Helmut Norpoth's model that got attention last week for audaciously putting the odds of a Trump presidency between 97%-99% was reverse-engineered to fit the data, a sort of complicated linear regression equation.
That's a reasonable approach, but it's not fool-proof. In fact, it's a big reason Nate Silver has so much egg on his face from his terrible predictions about the race so far. The problem with reverse-engineering formulas like this is that it doesn't take account of sui generis phenomena, and Trump is just that.
Specifically, Trump's self-funding (he's a real boy in a Republican field of pinnocchios); his pugnacious refusal to back down to anyone--the guy is getting nuked from outer space, attacked by his own party, by the president, by the pope--no one in the history of American politics has been able to withstand this kind of relentless onslaught and yet Trump isn't just withstanding it, he's thriving from it; and his strategic leveraging of social media (I tend to hear about the latest 'controversy' surrounding Trump from Trump himself before I see it in the major media) means that he can effectively fight the entire world on his own and come out on top. His position as king of social media gives him a wider effective reach than Fox News, the New York Times, and the WSJ combined.
The question still remains why this theoretical landslide isn't being detected in the polling data. I put little stock in hypothetical general election match-ups when both fields still have multiple candidates. The explanation as to why is tricky, but let's give it a shot.
When pollsters contact people, they filter out everyone who isn't a "likely voter" (in most polls anyway--some just ask about registered voters. LV is superior to RV). They then proceed to ask who these "likely voter" respondents would vote for if it were Hillary v Trump, Hillary v Cruz, Hillary v Rubio, Hillary v Kasich, Bernie v Trump, Bernie v Cruz, etc. Well, these respondents are now tagged as "likely voters" and so will usually answer even if the candidates in question won't motivate them to actually get out and vote a year later. These respondents weren't really lying when they identified as a "likely voter", but that likeliness may well be contingent on who the nominee is. In a race like this, that benefits Trump enormously.
An example of how this works in practice is you get a Trump supporter who is dead set on backing his, so when the pollster asks if he's a likely voter, he says he is. But if Trump's not the nominee, he's not actually going to go out and vote on election day. Theoretically this can go in any direction (Trump keeping Rubio supporters home), but given how the huge Republican surge thus far is overwhelmingly a consequence of Trump, it's far more likely that people will sit the election out if he's not the nominee than if he is.
The upshot of polls being conducted in this manner is that we get absurd results such as Kasich showing up as the best matchup for the GOP against Hillary in the general, while Cruz and Trump often appear to be the worst. But Kasich doesn't have any enthusiastic supporters, while Cruz and Trump do. Those Trump and Cruz supporters are still telling these pollsters that in Hillary v Kasich, they'll vote for Kasich, but in reality a lot of them won't.