Sunday, November 4, 2018

It's All Over...Except for Election Day's Votes in NC: An Analysis of 2018's NC Early Voters

Early voting has ended in North Carolina, and the numbers are beyond impressive.

Why "impressive"? Because the Old North State is in its 'blue-moon' election cycle for the 2018 mid-term. With North Carolina governor's races held in presidential years and neither of the two U.S. Senate races being contested this year, there isn't a major contest that draws substantial attention from across the state.

Yes, there is a weirdly contested election for a state supreme court seat, various seats on the lower courts, and controversial constitutional amendments on the ballot, but there's nothing like what is driving early voting in states like Texas (U.S. Senate), Georgia (governor's race), or Florida (both U.S. Senate and a governor's race).

So, to have over 2.1 million votes cast before 2018's Election Day, when the last mid-term election in 2014 saw a total of 2.8 million votes cast, allows for the term 'impressive' to be used in this context.

Now, we will continue to see a trickle of absentee by mail ballots come in, but with in-person absentee voting concluded, we have a sense of who has showed up to vote, but not necessarily 'how' they voted, until we get the early returns after the polls close at 7:30 (or when the last votes are cast) on Tuesday evening.

With that said, here are the dynamics of 2018's early voting in North Carolina (these figures represent all accepted absentee ballots, both mail and in-person, as of November 4). As a reminder: all data comes from the North Carolina State Board of Elections:

In comparison to 2014's mid-term election, 2018's mid-term election has well surpassed the numbers of early votes in the last mid-term election:

With the trickle of mail absentee ballots coming in over the next few days and that may catch up to 2014's total of 76,000 votes, 2018's final in-person absentee ballots are 79 percent above 2014's final in-person numbers.

The daily totals for absentee ballots saw the greatest number of in-person votes (nearly 200,000) being cast this past Friday, before early voting in-person ended on Saturday.

The cumulative daily totals for 2018's early votes:

In comparing 2018's early votes to 2010's and 2014's mid-term elections, along with the 2016's presidential election, 2018 has ended up closer to the presidential election two years ago, rather than to the past two mid-term elections.

In comparing the voter party registration to 2014 and 2016, registered Democrats achieved 67 percent of 2016's presidential year early votes, while registered unaffiliated voters achieved 68 percent of their 2016's early vote final numbers, and registered Republicans achieved 62 percent of their 2016 early vote final numbers.

In comparison to 2014, registered Democrats ended up 56 percent ahead of their final 2014 numbers, registered unaffiliated voters were 130 percent ahead, and registered Republicans were 65 percent ahead of their final 2014 numbers.

In looking at the final 2018 early vote by voter party registration, registered Democrats lead the party registration percentages:

To set this into some context, registered Democrats are 38 percent of the November 3rd NC voter registration pool of 7 million plus, so they ended up being +4 to their voter pool percentage. For registered Republicans, they equaled their voter pool percentage at 30 percent, while registered unaffiliated voters (the fastest growing group among NC registered voters) was 27 percent of the state's early votes, down from their voter pool percentage of 31 percent, or -4.

One thing to note, however: you can't necessarily make the claim that since registered Democrats are overperforming, all of these Democratic registered voters will be party loyalists. North Carolina's white, older, and rural voters tend to be more conservative, and thus are actually Republican voters. But we'll find out come Tuesday night, perhaps.

In comparing the daily totals and cumulative totals by voter party registration:

The last day of early voting, while not the highest number of votes cast (due to early voting ending at 1 PM on Saturday), saw an uptick in registered Democrats, while both registered Republicans and unaffiliated voters tied at 28 percent each.

To give some further context to how each party registration group ended in comparison to their overall voter registration pool numbers, here is the turnout based on party registration:

Nearly 30 percent of the 7 million plus registered voters cast their ballots in early voting, either through mail or in-person. Registered Democrats had a slightly higher turnout so far, versus registered Republicans (slightly above the state average), while not quite a quarter of registered unaffiliated voters turned out.

Next, the analysis turns to several demographic characteristics of North Carolina's early voters, first by gender of voters and their party registrations:

Women are slightly ahead of their voter registration pool percentage of 53 percent, and saw a higher percentage of registered Democrats casting ballots than their voter pool percentage (42 percent), which is shown in the next chart:

Men also saw a slightly higher Democratic party registration percentage (36 percent) show up than in their voter pool percentage (34 percent).

Next, in looking at the regional distribution of NC's votes within four distinct areas: central city voters in urban counties; suburban voters outside of central cities but within urban counties; surrounding suburban county voters; and, rural county voters. See here for specifics of the coding for these regions.

A majority of the early votes are from suburban voters, both those within urban counties (but outside the central city) and in the surrounding suburban counties within the Metropolitan Statistical Area counties. Based on voter party registration, the two different suburban regions are almost mirror images of each other, in terms of party registration of voters who showed up early.

For comparison, here are the voter party registration percentages within each region from the 11-3-18 voter pool:

In comparison to their voter pool registration percentages, suburban registered Democrats (both within the urban county and within the surrounding counties) have overperformed their numbers. Registered Republicans are slightly below their voter pool in surrounding suburban counties.

Next, I combined both gender and region to create analysis of women and men within regions and their party registration percentages:

If this is the year of the suburban voter, then these numbers trend for a healthy showing of suburbanites in the final electorate, for both men and women.

Next, in looking at voter race and party registration for North Carolina's early voters:

And their comparison to the 11-3-18 voter pool percentages by race and party registration:

White voters overperformed their voter pool percentage by +3, while black/African-Americans were equal to their voter pool percentage (22 percent). While white registered Republicans were even with their voter pool percentage, white registered Democrats were above their voter pool percentage (+4).

The next two charts show the daily numbers and daily percentages by voter race:

Black voters took advantage of both Sunday's "Souls to the Polls" efforts and also the last Saturday of early voting, with a higher percentage showing up to cast ballots.

The final demographic aspect to 2018's early voters are by generational cohorts, as defined by the Pew Research Center's work on generations:

And each generation's cohort within the potential electorate (voter registration as of Nov. 3), as compared to their early vote percentages:

For all the talk of mobilizing young voters, most often ascribed to Generation Z and Millennials, their performance in NC's early voting was lower than their voter registration percentages. Early voters were an average of 56 years old, with Baby Boomers and older voters making up the majority of early ballots accepted.

Here's a histogram (distribution) of North Carolina early voters by age:

For comparison, here's the histogram of the remaining voters who didn't vote early in North Carolina (with a mean age of 46 years old for voters left in the pool who could cast votes on Election Day):

To give some further sense and analysis to how generations made up past early vote electorates in North Carolina, the next three charts give the daily percentages of early voters by their generational cohorts:

As another example of how 2018's mid-term is not quite a mid-term but not quite a presidential year, I looked at Millennials in 2014 and 2016 and Millennials/Generation Z for their daily percentages of casting early votes:

2018's mid-term shows younger voters slightly closer to 2016's daily percentages, but not quite there.

One of the more interesting aspects to this year's early voters is how they participated (or didn't) in the last mid-term election of 2014. Here are the numbers by vote method and registration status of 2018's NC early voters:

The percentages of the 2018 early voters and their 2014 vote methods & participation are:

  • 2018 early voters who cast their vote absentee in-person in 2014: 35 percent
  • 2018 early voters who cast their 2014 vote on Election Day: 27 percent
  • 2018 early voters who were registered before/in 2014 but didn't vote in 2014: 16 percent
  • 2018 early voters who were registered after 2014: 19 percent

Based on voter party registration, I looked internally into the top six categories in the above chart, by both raw numbers and percentages:

Finally, I looked at an analysis of within each party registration and how their early voters cast ballots in 2014 or didn't:

For registered Republicans, 30 percent of 2018 early voters cast their ballots in 2014 on Election Day, while 26 percent and 25 percent of registered Democrats and unaffiliated voters (respectively) voted on Election Day four years ago.

With over a quarter of NC's early voters having been Election Day voters on 2014, the question becomes: who will likely show up on Election Day Tuesday?

In looking at the 7 million plus registered voters, one can look at each voter's voting history and determine who would be likely voters to cast ballots. If a voter has cast a ballot in each general election from 2008 to 2016, I labeled those voters as "high propensity" voters. A second category would be "presidential year" voters, those who cast ballots in 2008, 2012, and 2016, but didn't vote in 2010 or 2014 mid-term election years. These "presidential year" voters could be motivated to cast a ballot, whereas it is likely that the "high propensity" voters will continue their voting streak and cast a ballot.

In using these two categories, I looked at who has cast ballots in early voting and who was still left in the voter registration pool, using the voter's party registration to look inside each parties potential 'electorate':

Registered Democrats have had 60 percent of all of their 'high propensity' voters cast an early ballot, whereas 56 percent of all 'high propensity' registered unaffiliated voters cast an early ballot, and 51 percent of all 'high propensity' registered Republicans voted early. 

Of the presidential year voters (those who voted in presidential years 2008-2016 but not in the 2010 or 2014 mid-terms) in each party registration, about a quarter of each party's presidential year voters cast an early ballot in 2018. 

Then, I looked at each party's high propensity voters, combined with their presidential year voters, and looked at the totals of voters in each party registration by if they had cast an early ballot, or potentially were awaiting Election Day 2018. Here are the raw numbers and the percentages within each party registration:

Over half of high propensity/presidential year registered Democrats cast an early ballot, whereas 48 percent of high propensity/presidential year registered unaffiliated voters and 46 percent of high propensity/presidential year registered Republicans used early voting to cast their ballots.

Finally, to give folks a comparative analysis and sense of how North Carolina voting methods tend to break down, by voter party registration (the striped bars) and by the final results (the solid bars) within the election contests, here's 2014's U.S. Senate election (a competitive race between Republican Thom Tillis, the winner, and then incumbent Democrat Kay Hagan) and the state's 2016 presidential battle:

I'll have another blog post before Tuesday, focused on North Carolina and how we might look at the returns for congressional and state legislative races, and which ones to watch in particular, based on election fundamentals.