New York Times (North Carolina):
By Nate Cohn
October 26, 2016
Our estimate of the final vote, based on what we know about the early vote
There aren’t many states more important this November than North Carolina, a rapidly changing state crucial to Donald J. Trump’s hopes of winning the White House and to Democrats’ hopes of winning the Senate.
There also aren’t many states with better election data than North Carolina. The state releases detailed, individual-level information on every voter in the state. It even publishes a daily account of who has voted early, either in person or by mail.
To get a better sense of what’s going on, we’re trying an experiment: We’re estimating the result of the early vote and the Election Day vote by combining the rich data released by North Carolina with data from the Upshot/Siena survey of North Carolina.
Already, about 812,000 people have voted in North Carolina, out of about 4,425,000 we think will eventually vote. Based on the voting history and demographic characteristics of those people, we think Hillary Clinton leads in North Carolina by about 6 percentage points. We think she has an even larger lead – 22 percentage points – among people who have already voted.
Here’s a breakdown of those estimates for the two major-party candidates:
These aren’t official results. They’re just estimates. If our polling is wrong, then our conclusions could be wrong as well. Our polling has been somewhat stronger for Mrs. Clinton than other surveys of the state have. The last Upshot/Siena poll this week gave Mrs. Clinton a seven-point lead.
Even if our polling is pretty good, there’s still uncertainty: Undecided voters and the supporters of Gary Johnson pose as much of a challenge for this project as they do for a typical survey. But by the end of the early vote, one of the biggest sources of uncertainty in polling – the composition of the electorate – will be greatly diminished.
As early voting continues, our estimates will gradually shift as low-turnout voters lock in their votes, and as doubts arise about whether people who haven’t voted yet actually will.
The big thing to watch is the trend: Is the early vote changing the composition of the electorate to the advantage of either candidate?
How many people have voted, and how many we think are yet to vote
And how our estimates of the final vote have changed
Heading into the early vote, our estimate was that Mrs. Clinton had a six-point lead in the state. The early vote so far hasn’t been enough to move that assessment. That’s because only a fraction of voters have turned out. It’s also because most of the people who have voted so far are fairly reliable early voters.
What would cause it to change? The easiest way to shift these estimates is if one candidate benefits from a strong turnout among voters whom we didn’t expect to vote, especially people who weren’t registered to vote at the start of the early voting. In general, low-turnout voters tend to vote closer to the election.
Two things have changed, albeit slightly. First, our estimate for the final turnout has gradually declined. That’s because early voting has been a little slower this year than in 2012. Part of the reason is that there are some North Carolina counties where the number of in-person early voting stations has been scaled back. This has clearly reduced the number of early voters. We have made no adjustment for this effect, which should gradually diminish once more polling stations open on Oct. 27.
The other change is that the supporters of Mr. Johnson simply aren’t turning out in early voting, according to our polling data. So far, he has just 3 percent of the vote among people who voted early, according to our polling and our modeled data.
Below, our estimates of the vote in North Carolina across different groups of voters.
By race
From the roughly 812,000 early votes:
From the roughly 3,612,000 votes yet to be cast:
Our best guess at the final vote in North Carolina:
At the start of early voting, we estimated that 71 percent of the final overall total of voters would be white and that 21.4 percent would be non-Hispanic black.
These pre-early-voting estimates were strictly based on vote history. A 45-year-old white person who voted only in 2012 and a 45-year-old black person who voted only in 2012 were given an equal chance of voting in our forecasting.
In practice, vote history isn’t the only factor that determines turnout. In recent elections, black voters in North Carolina have been likelier to vote than white voters with the same vote history. For this particular project, our estimates start out agnostic on whether black or white voters will turn out in greater numbers than their vote history suggests. The turnout should speak for itself.
If early voting goes as it did in 2012, our estimate for the black share of the electorate will gradually increase if infrequent and even previously unregistered black voters show up in disproportionate numbers.
Over all in 2012, black voters represented 27.3 percent of the early vote and ultimately represented 23 percent of the final electorate, even higher than their share of registered voters (22.4 percent).
By vote history
From the roughly 812,000 early votes:
From the roughly 3,612,000 votes yet to be cast:
Our best guess at the final vote in North Carolina:
Every vote counts, but not every vote tells us as much about who is winning the early vote.
There are some people who are all but sure to vote: They participate in just about every election. If they vote early, it doesn’t really change our view of the race. But if a voter whom we don’t expect to vote shows up, that moves the numbers. That’s what the campaigns are trying to do, too: They’re trying to get their less likely supporters to the polls.
One easy way to get a sense of whether infrequent voters are turning out is to look at whether they participated in the 2014 midterm election. Our estimate heading into the early vote was that voters who skipped the midterm election would make up 42 percent of the electorate, and we thought they would support Mrs. Clinton by a 10-point margin.
It’s worth watching both parts:
* Whether Mrs. Clinton’s expected 10-point margin with that group grows or shrinks, a sign that the infrequent voters who are turning out are more or less favorable to her than expected.
* Whether they’re expected to make up more or less than 42 percent of the electorate, a sign that more of them have stayed home or turned out than expected.
By age
From the roughly 812,000 early votes:
From the roughly 3,612,000 votes yet to be cast:
Our best guess at the final vote in North Carolina:
So far, most early voters have been pretty old. Young voters were most likely to be undecided or in support of Mr. Johnson in the Upshot/Siena poll, so Mrs. Clinton doesn’t benefit as much from a high youth turnout as you might guess.
By party
From the roughly 812,000 early votes:
From the roughly 3,612,000 votes yet to be cast:
Our best guess at the final vote in North Carolina:
Democrats have a longstanding voter registration advantage in North Carolina, but a significant slice of them are conservative, older white Democrats who have been voting Republican in presidential elections.
There isn’t a realistic scenario in which registered Republicans would outnumber registered Democrats in the final count. That’s especially true in early voting, which is traditionally used more by Democrats than Republicans.
In 2012, Democrats had an edge of 48 percent to 32 percent in party registration among early voters, and a 44-33 edge in the final count. Since then, the Democratic registration edge statewide has diminished, in large part as older conservatives have switched to the Republicans. At the same time, newly registered voters who support Democrats have been far likelier to register without affiliating with a party.
As a result, the expected Democratic registration edge is somewhat smaller than in the past: 40 percent to 32 percent for the Democrats.
For more information, go to: www.beverlyhillsimmigrationlaw.com
No comments:
Post a Comment