There's now enough evidence in the early voting numbers to suggest that Republican efforts to suppress voters of color in a post Voting Rights Act-era are working exactly as intended.
In the figure below, we present each group’s share of the early vote compared to their size of the overall CVAP. This tells us whether a group is “overperforming” or “underperforming” in early voting relative to their presence in a state’s population. For instance, if Latinos make up 15 percent of eligible voters but only 10 percent of the early voting population, they are underperforming in the early vote by 5 points.
Using this perspective, there is only marginal evidence that the surge in Latino turnout in 2016 is resulting in an outsize Latino electorate. Florida is the only state where Latino early voting is outpacing both population growth and growth in the early vote for non-Latinos.
Even there, the growth in participation is relatively small, and Latinos still make up a smaller proportion of early voters than they do of the eligible electorate. So far in 2016, the Latino share of early vote in Florida is 5.1 percentage points lower than the Latino share of the eligible electorate.
In two other states with large Hispanic populations — Texas and Nevada — Latinos are underperforming relative to population growth and steady or increasing early voting by whites. Although many of these white voters may be Democrats, this suggests a more nuanced story about the likelihood that high Latino early voting alone will reshape the electoral landscape.
At the same time, early voting among African Americans is lagging behind their share of the eligible electorate, often in dramatic ways.
In 2012, African Americans overperformed in states such as North Carolina and Texas. But this year, black voting rates are trailing other groups relative to their size of the electorate, with some swings on the order of 5 to 10 percentage points.
Even in North Carolina, where the black share of the early vote is on par with their share of the eligible electorate (and similar to white voters), the pattern is a departure from 2012, when African Americans made up 5.8 points more of the early vote than their population share.
Meanwhile, white voters constitute a larger share of the early vote than in the past two election cycles. For example, in North Carolina and Florida, whites were underperforming in the early vote a week before the 2012 election. But they are now overperforming — and substantially so in Florida.
Of course, it is not entirely clear what this suggests for Tuesday’s outcome.
It does suggest that overall turnout in early voting states is up. That's a good sign for Democrats. But in all six of the above states, it's white voters who are turning out at a higher percentage. That's better news for the GOP. If these white voters are college-educated voters that Clinton would be winning for the first time in decades, then it's good news for her. If these are disaffected white Trump voters driven to the polls by anger, who never really bothered to vote before, then it's big for him.
Either way these are the votes that the polling models and turnout scenarios have missed in the past.
We'll see what happens here in the final week.