1. We've seen a lot of discussion of lower African-American turnout this year compared to 2012. There's been speculation that that is tied to Barack Obama not being on the ballot. But now that early voting is coming to an end we can see that's not the real story. In most cases, African-American turnout in the early vote is actually up - except in the places where state legislatures took the most aggressively steps to limit African-American voting. Most specifically in North Carolina. This is where a federal judge said that the state had specifically designed its new voting laws to limit African-American voting. These are neo-Jim Crow laws. And it's had an effect.
Michael McDonald is one the top academic political scientists working in the field of election data.
Something went very wrong for African-Americans' voting in North Carolina pic.twitter.com/ZpwjyEavmd
— Michael McDonald (@ElectProject) November 6, 2016
It is important to note that people who were early voters can become election day voters and vice versa. So it's certainly possible that the picture could look significantly different when all the votes are counted. It's also true that these numbers have gone up in percentage terms over the course of the voting. They could go up more. Still, there's an obvious difference between these states. And the most obvious explanation is the bundle of voting restrictions and purges in North Carolina.
2: One of the reasons it's hard to draw firm conclusions about early voting is that there is a steady growth of early voting from election to election. Everybody's share of the early vote is going up, or almost everybody's. Much of that is that people are shifting from election day to early voting. So changes in the early vote are not necessarily measures of increasing participation or enthusiasm. We can still draw inferences, in some cases very strong inferences. But that's why a lot of the conclusions are tentative.
3. The news about a surge of Hispanic voter participation has now broken way out of the numbers nerd universe is hitting the front pages of the news. I've been cautious about making too many assumptions about what's happening for the reasons I noted in item two. But it now seems clear that something really big is happening. The situation is Nevada seems pretty clear. The situation in Florida looks increasingly clear. One of the key issues we discussed above is how much each parties' growth of early voters is just cannibalizing their election day vote - just moving voters around rather than adding voters. One key data point. Daniel Smith looked at the numbers through Saturday. He found that Republicans were cannibalizing their same day vote at a slightly higher rate than Democrats. Smith also notes that of the 907k Hispanics who had already voted by Saturday, 36% didn't vote at all in 2012. That is 12 percentage points higher than whites. In other words, a lot of the Hispanic early vote is made up of either new voters or people who didn't vote in 2012. That could be because they just turned 18, have recently become citizens or have just become more politically engaged. In any case, those are big numbers.
4. This may be obvious. But another key issue with the growing Hispanic vote is that the public polls seem to have missed it or at least missed it a significant amount of it. We know that has boosted Clinton a lot in Nevada. It has probably done the same in Colorado and Florida. It may even put Arizona in play. We've heard a lot about poll failure and 'missing white voters'. It seems like we are seeing a genuine poll failure, albeit a relatively minor one. But it's Hispanic voters who seem to have been missed. It's something of an open secret that public polls have hard time accessing Hispanic voters. There are higher percentages of cell phone only voters. There are significant percentages who may prefer to speak in Spanish rather than English. There are ways to get around these challenges. Polls which focus exclusively on Hispanic voters do much better. But we have a lot of state polls and lower quality digital and robopolls that don't do these things. So it's not a huge surprise that this could have been missed. But it could have a significant impact on the results on Tuesday.