Exploring Evangelicals in 2016's Election: Partisanship Divided The Flock

I found this article on "A Quiet Exodus: Why Blacks Are Leaving White Evangelical Churches" to be an interesting exploration of the controversies surrounding evangelicalism in today's political environment.

In particular, Michael Emerson, co-author of "Divided by Faith", a study that explored the racial relations within the evangelical movement, said in the article:

"Everything we tried is not working. ... The election itself was the single most harmful event to the whole movement of (racial) reconciliation in at least the past 30 years,” he said. “It’s about to completely break apart.”

Much has been made about the support by most white evangelical voters to President Trump. In fact, one of the president's evangelical advisory board members said recently:

Generational Partisanship (or Why Millennials May Not Save Us from Partisan Loyalty and Polarization)

As many may have realized by now (by reading some of my previous posts), I've become very interested in the generational dynamics underway in American politics, and particularly in North Carolina (through voter registration figures and analyses). But voter registration only tells part of the story--a companion aspect to our political environment is voting "behavior," especially about voters' party self-identification and supporting presidential candidates, when it can be a driving force of our partisanship and polarization.

In trying to better understand the partisanship and potential polarization coming from party identification in our nation, along with the generational dynamics that are happening (especially with the rising Millennial generation), I ran a series of analyses using survey data from the American National Elections Studies (ANES) to investigate the following questions:

• Are Americans really moving into political independence with the "rise of the independent" or are we seeing another phenomenon, especially when it comes to voting behavior of these independents?
• What have been the trends in recent presidential elections when it comes to partisan self-identification and presidential vote choices among the four generations in the American electorate?

Analysis of Generations in the 2016 Election: Partisanship, Ideology, Vote Choice, and Feelings Toward Clinton & Trump

In exploring more about the "generational shift" occurring in North Carolina politics (particularly from a voter registration and turnout approach--see previous posts on this blog site), I wanted to take a larger view of the differences between Millennials (those born after 1980) and Baby Boomers (those born between 1945 and 1965), with Generation Xers (born between 1965 and 1980) and the Greatest Generation (born before 1945).

One invaluable data resource to explore these differences is the 2016 American National Election Study, a massive survey of over 4,000 individuals who are contacted both before and after a presidential election. This survey asks a battery of questions, not just on politics but a variety of policy questions and a range of issues.

For those who would like to use the 2016 ANES data, the folks at UC-Berkeley have developed a website that allows users to put in various questions (by first finding the variable) from the survey and run cross-tabs to analyze the survey pool (it's a little tricky at first, but once you get the hang of it, you can spend a great deal of time running numbers).

For example, if you wanted to "recode" the "age" question of the respondents into generations (in 2016, Millennials were 18-35; Generation Xers were 36-51; Baby Boomers were 52-71; and Greatest Generation members were 72-90), you would find that the 2016 pool ("electorate") broke down into the following generational cohorts:

Using this division of the 2016 ANES data, I began to run cross-tabs of the data by different characteristics and policy questions asked in the survey. For example, what was the party self-identification by the generations?

Democrats (those who identify as strong, not strong, and leaners combined) are a plurality of both Millennials and Gen Xers, while Baby Boomers have a nearly even divide between the two political parties. Members of the Greatest Generation have a plurality of Republican self-identification. Barely a third of Millennials self-identify with the GOP, while nearly 18 percent of the youngest generation don't identify with either political party.

Along with partisan identification is ideological self-identification; the first analysis that I ran included the response for "haven't thought about it" to reflect possible uncertainty of one's ideological perspective (conservative, liberal, moderate, or haven't thought about it):

Not surprising, Millennials have the largest portion within a cohort that "haven't thought about it," while the oldest voters (Baby Boomers and the Greatest) have the lowest portions within their cohorts of "undecided" ideological identification.

Dropping the "undecideds/unknowns" out of the analysis paints an ideological picture of the four cohorts:

Conservatives claim a plurality of Baby Boomers (43.8 percent) and a majority of the Greatest Generation (50.9 percent), while Gen Xers are nearly evenly divided amongst the three categories. A plurality of Millennials, as documented in other surveys such as the Pew Research Center, are liberal in their ideology. But will Millennials potentially grow "more conservative" to match their older cohorts when they reach 50 years and older?

To compare, I went to the 2004 ANES Survey and, adjusting for the age categories, ran a comparison of ideological self-identification between the two election surveys.

While the oldest Millennials in 2004 were only 23 years old, as compared to 35 years old in 2016, there was an increase of Millennials who identified as conservative over the twelve year period, at the expense of those who identified as "moderate;" liberal identification remained fairly stable. Among the Greatest Generation, however, a substantial increase (10 points) in conservative ideological self-identification occurred between 2004 and 2016, again at the expense of those who identified as moderates.

Having looked at both the partisan and ideological self-identification of the generations, I looked at the presidential vote choice among the cohorts:

The almost mirror images among Millennials and Greatest generation cohorts is pretty striking: while nearly 55 percent of Millennials cast their ballots for Clinton, nearly 58 percent of the Greatest Generation went for Trump. While there was a 17 percentage vote gap between Greatest Generation voters for the two candidates, the 20 percentage vote gap among Millennials is significant, taking into account the likelihood that they are more liberal and more Democratic in identification.

The greatest vote share among the cohorts going to third-party candidates was, not surprisingly, among Millennial voters.

Much was made about the unpopularity of both major party's candidates. In looking at two questions about the presidential candidates and the respondents' feelings about the candidates, another generational divide emerged: first, as demonstrated in the combined "feelings thermometers" when asked about both Clinton and Trump.

A feeling thermometer asks a respondent, on a scale of zero to 100 (zero being ice cold, 100 being burning hot), to state a number as to how they felt about each candidate. In classifying "warmth" as being any response over 55 and "cool" as being any response under 45, I grouped each generation's responses to both candidates into the following "warmth" and "coolness" towards both Trump and Clinton.

While all four groups saw consistently similar responses to Clinton, typically at 50 to 53 "warmth" and 40 "cool," it was the generational differences in responding to Trump that stands out: while Millennials were barely more "warm" than "cool" to the GOP candidate, the 30 to 40 percentage gap among warmth and coolness by Baby Boomers and the Greatest Generation is striking.

Another set of questions asked about whether the respondents were "never proud of" either candidate or were "always disgusted by" a candidate.

While all generations were fairly consistent in their responses to the questions regarding "never proud of" or "always disgusted by" Hillary Clinton, the generational differences among the responses to Donald Trump shows that Millennials were the generation showing the greatest level of loss of pride in Trump and "always disgusted by" the GOP candidate.

But even with this level of disgust by Millennials, it clearly did not translate into enthusiasm in their voting:

Millennials had the lowest level of "strong preference" for their presidential candidate among the cohorts, while the Greatest Generation had three-fourths of respondents express strong preference for their respective candidate.

One of the more interesting findings regarding voting was a question posed in the ANES regarding the election integrity: does the respondent believe that the votes are being counted fairly?

For all of the talk about vote and ballot integrity, it was Millennials who had the lowest outlook on whether their vote would be counted fairly:

To wrap up, much of this ANES data is in line with another major study of the generational divide in American politics from the Pew Research Center. While older voters have higher participation rates than younger voters, and thus dominate the electoral fortunes, the tectonic shift that is currently underway in the country will reshape the political landscape.

In the next blog post, I'll be looking at this generational divide through the lenses of various policy issues and concerns, such as the state of the 2016 economy, transgender bathroom policy, the ACA's (Obamacare's) effect on health care costs, marijuana policy, and the role of government.

The Final NC Voting Data Has Been Released

Well, I know it's been a while since I've posted, but I needed some time away (and time to get caught up on my other jobs), but now that the North Carolina State Board of Elections has released some final data on voters, we can truly begin to dissect the 2016 electorate in the Old North State.

Much speculation was made leading up to the November 8 general election about what this year's NC electorate would look like, and with the dynamics of early voting presenting a whole new dynamic in the state, it was anyone's guess what the final picture would look like in the state.

North Carolina on Election Day: Some Basic Information & Political Dynamics (in Charts, of course)

Well, it's finally, finally, FINALLY here.

Just to give a sense of what North Carolina has seen over the past few general elections in presidential election years (of course, by charts):

Here are the presidential election results for North Carolina:

Here are the 2008 and 2012 exit polls for the state.

Here is a chart documenting the North Carolina registered voter pool from 2002-2016:

Here are the registered voter pools and the electorates in general elections by the percentages of party registration in North Carolina:

Ideology of North Carolina (as expressed in exit polls), 1996 to 2012:

North Carolina general elections by voting method: absentee (early) ballots versus election day ballots cast, 2004-2012:

Here are the turnouts based on registered voter groups in North Carolina, first by race:

The turnouts based on registered voter groups in North Carolina, by party registration:

Next, the turnout rates of North Carolina registered voters by age (note: these age bands are determined by the NC State Board of Elections):

Next, the composition of North Carolina general election electorates by race:

The composition of North Carolina general election electorates by party registration:

The composition of the North Carolina general election electorates by age categories:

2012 presidential election margin of victories in the 100 North Carolina counties with 13 counties (marked in yellow) delivering 50 percent of the state's votes:

Many of North Carolina's counties are becoming very polarized. The NC counties by partisan competitiveness or 'landslide' in margins of victories:

And then the North Carolina vote totals in these competitive/landslide counties:

I'll be posting more charts and data about North Carolina as Election Day rolls on. I'll be an election analyst for WFAE, Charlotte's NPR Station, starting at 7 PM this evening. Thank you all for following this blog; it has been a tremendous ride. Michael

UDPATE: I ran this analysis back after the March 2016 presidential primary in North Carolina, and I think it speaks to the potential support for Donald Trump in this general election. I ran a number of different 'factors' against the county performance for Trump, and the most 'influential' factor was the county's percentage of college-degree holders:

Nearly 50 percent of Trump's 2016 NC county presidential primary support can be explained by the college's percentage of those who hold Bachelor's degrees or higher, and that is an inverse relationship (meaning, the lower the county's percentage of holding a Bachelor's degree, the higher Trump's support was).

UPDATE at 4:41 PM: above were some 'competitive/landslide' maps and data on North Carolina's 100 counties. In current state politics, the divide between urban and rural areas of the state has grown more pronounced, but suburban areas tend to be the most Republican areas in the state, based on total numbers cast for presidential candidates in 2012 in the three areas:

For Democrats, breaking into the spread in suburban North Carolina and making sure the urban areas return large numbers (and perhaps spread the difference out more) will be key to winning in the state.  With three-quarters of the votes come from urban & suburban counties, these will be the key areas to be watching tonight.

Differences in Absentee Ballots: 2012 vs. 2016 by Party Registration in North Carolina

With the dramatic increases in absentee ballots in North Carolina and preparing for the general election day, a question from a Twitter follower was, "what areas saw the most changes among absentee voters?"

Using information on accepted absentee ballots in both 2012 and this year (with some changes in the numbers still expected), I was able to run cross-tabs of the different party registration based on two key factors: region and race.

First, the Democrats saw a very slight drop in their absentee ballots (both mail-in and in-person) from 2012's totals. In looking at the cross-tabs by region (urban, suburban, and rural counties) by race, the following demonstrates the changes in the Democratic voter base among absentee voters:

Among white voters, Democrats overall saw a 5 percent increase in registered members of their party showing up to cast absentee ballots, but there were noticeable differences by region: rural white Democrats were slightly down, while urban and most notably suburban white Democrats were up over their numbers from four years ago.

The critical change from 2012, to the negative, is among black Democrats. Only among suburban black Democrats was there an increase from four years ago, while among urban and rural black Democrats, the drops were noticeable to have an overall drop of 8 percent from 2012's numbers. 

Among all other races, however, registered Democrats saw, at times, dramatic increases in their numbers from 2012's totals. And while some of the changes were due to what would appear to be small numerical changes (for example, Indian Americans went from 7,372 to 7,485), registered Asian Democrats went from 8,146 to 12,345, while those voters who classify themselves as 'other races' went from 17,220 to 24,677.

Overall, registered Democrats saw their greatest decline in their numbers from rural counties, down 10 percent (from 324,373 to 290,856) and contributing to the slight decrease in their overall numbers. Yet suburban Democrats, going from 200,445 to 215,107, showed the greatest increase. 

Among registered Republicans, the total numbers went up by 13 percent, but there were slight differences among the regions and racial categories. 

While white Republicans were up 12 percent, it was suburban white Republicans who were up the largest in their numbers (up 30 percent), while white urban Republicans were up on 4 percent. 

While the black Republican numbers seem dramatic, it was an overall difference between 10,949 in 2012 and 10,067 this year in terms of ballots being cast.

Again, the other races were among the consistent growth areas, but like with black Republicans, the numbers were fairly small: Asian Republicans from 3,624 to 4,865; Indian Americans from 1,937 to 2,776; and multi-racial Republican voters from 1,604 to 1,985. Among those voters who did not indicate a race on their voter registration form, the noticeable increases was from a total of 9,783 in 2012 to 17,029. Overall, like with registered Democrats, suburban registered Republicans saw the greatest overall regional increase, of 31 percent. 

Finally, the huge increase of registered unaffiliated voters is dramatic in comparison to their partisan counterparts. 

White registered unaffiliated voters saw an overall increase of 41 percent, but among suburban white registered unaffiliated voters, it was an increase of 57 percent. 

Among black unaffiliated voters, who are typically more from the Millennial generation, it was again suburban black unaffiliated voters showing the largest increase. 

Again, like the partisans, all other races saw huge increases among unaffiliated voters. Among Asian registered unaffiliated voters, their numbers in 2012 was 10,422; this year, there was 17,340. The other numerical large-scale increase was among 'other' races, which went from 12,850 four years ago to 20,670 this year. 

The accepted absentee ballots' electorate could be anywhere from 60 to 65 percent of the total electorate (ballots cast); in 2012, 61 percent of the ballots in that election came before Election Day (56 percent from in-person absentee ballots, with the remaining 5 percent from mail-in absentee ballots). If the overall electorate increases beyond 4.5 million ballots (as it was in 2012), then the Election Day electorate will be critical to watch as the absentee ballots may be on the lower end of the 60 percent range. 

Some Exploration of the True Wildcard in NC's Early Voting: Registered Unaffiliated Voters

With the near-final numbers posted in terms of North Carolina's absentee voting before Tuesday's election, the true wildcard seems to be the huge spike in registered unaffiliated voters who have cast a quarter of all the absentee ballots.

Their numbers are up over 40 percent from where they were in 2012 at the end of the early voting period, so I decided to look more closely at these voters who could potentially hold the balance for which party wins in North Carolina.

First, a general overview of the state's unaffiliated voters: where registered partisans (Democrats and Republicans) in North Carolina tend to be older, registered unaffiliated voters skew younger, due to the growing influence of Millennial voters; nearly 40 percent of registered Millennials are unaffiliated.

We see this influence of younger voters in breaking down each party registration category. Voters under the age of 51 are a majority (55 percent) of the registered unaffiliated voters who have cast accepted absentee ballots so far in this year's election in North Carolina.

One of the key dynamics going on in North Carolina's voter pool is the fact that most Millennial voters (who are the second largest generation cohort in the state's voter pool) live in urban counties; these urban areas also dominate the state's politics and voter pool as well.

With urban counties making up 54 percent of the state's voter pool, and Millennials at 30 percent of the voter pool, this combination is shifting the dynamics of the state.

I next isolated all of the state's registered Millennial voters and looked at their location and party registration.

With 61 percent of Millennials living in urban counties, they tend to split between registering as Democrats and unaffiliated, with 20 percent registering as Republicans.  In the other areas, registered Millennials in North Carolina tend to be much more unaffiliated than partisan. 

So, this gives us a sense of Millennials and their influence in North Carolina politics.

Now, we turn back to the unaffiliated absentee voters who may be the key to this year's election in North Carolina. 

One crude 'proxy' that might get at how these unaffiliated voters are potentially casting their ballots for the general election would be if these unaffiliated voters cast ballots in the March 2016 primary in North Carolina.

For unaffiliated voters, they have to pick a party to cast a ballot in for the primary election, and we can match their March 2016 voting records of which ballot they requested with their current status. Granted, this 'proxy' may not necessarily capture all aspects of the dynamics of a voter, simply due to the partisan nature of many North Carolina counties. To have any influence in a county's politics, especially if you are in the minority party, you may have to be a registered unaffiliated voter and cast your primary vote in the party opposite your true partisanship, and then return home in the general election. Or perhaps a registered unaffiliated voted in a party's primary to vote against a particular candidate; again, this proxy may not be the best one out there, other than asking "for whom did you cast your general election ballot for?"

With those caveats said, among the 807,000 unaffiliated voters who have cast accepted absentee ballots so far, the overall composition of this group by generation and region point to a distinct slant:

Among the early unaffiliated voters, 38 percent are urban Millennials and Generation Xers (under the age of 51), with another 19 percent being urban Baby Boomers. North Carolina's urban areas have been trending more and more Democratic in nature, as evidence by Obama receiving 56 percent of the urban county presidential vote in 2012. 

We also know that Millennials tend to be more Democratic-leaning in their partisan affiliation. 

In using the 'proxy' of voting in North Carolina's March primary, the thinking is that we could see a sense of where unaffiliated voters may 'see' themselves as partisans. 

Of the 807,000 unaffiliated voters who have cast accepted absentee ballots so far for the November election, 44 percent (348,000) cast ballots in one of the two major parties' primaries in March of 2016. 

Of that 348,000 primary voters, 56 percent cast ballots in the Republican primary, while 44 percent cast them in the Democratic primary.

So, one would think that NC's unaffiliated voters are perhaps more leaners to the Republican Party than the Democratic Party. 

But in looking at these possible lean-unaffiliated voters, a plurality (47 percent) of the GOP-primary voters and 67 percent of the Democratic-primary voters in March were urban county voters.  In total, of the 348,000 March primary voters, 56 percent of them were urban county voters. 

Among those urban county unaffiliated voters, Millennials and Generation Xers made up 50 percent of the ballots being cast in the two parties. And of those two generations, which party did they cast their ballots in the primary?

So in urban counties, which have dominate the November accepted absentee vote totals so far with 55 percent of the accepted ballots cast, urban unaffiliated voters who are under the age of 51 and who cast ballots in the March primary election did so towards the Democratic primary (63 percent) over the Republican primary (37 percent). 

Granted, half of the urban county unaffiliated voters who voted in March split between those over and under the age of 51, but making a claim that a majority of unaffiliated voters are more Republican-leaning than Democratic may be masked by the voter's location and a more detailed analysis of which party they participated in for the March primary.

Ultimately, this may be an exercise in probability, simply because of the 807,000 unaffiliated voters who have cast accepted absentee ballots for Tuesday, the majority--57 percent--of them didn't participate in the March primary and we have no idea of how they may lean in their partisan affiliation. And of these non-March primary voters, over one-third of them (36%) are Millennial voters.

Based on the possible different interpretations of the above data, we should probably continue with the belief that this pool of 800,000 unaffiliated voters, along with another chunk of unaffiliated voters on Tuesday, will tip the balance of partisan power one way or the other in North Carolina. There's just too much uncertainty about this influential and unknown group of voters to make a call one way or the other, but different ways of looking at them can produce potential benefits or disadvantages to one party or the other. 

North Carolina Absentee Ballots Near 2.2 Million

With another nearly 225,000 votes cast on November 1, North Carolina total absentee ballots have reached nearly 2.2 million, with nearly 2 million of those coming by in-person absentee ballots. Due to a busy work day, here's an abbreviated version of the post; more analysis to come later this afternoon.

This represents a nearly 10 percent increase over the same day in 2012 total absentee ballots.

The party registration break down among the two different absentee ballot types--mail-in and in-person--shows Democrats continue to have the overall lead among all ballots. Today, registered Democrats are 43 percent, registered Republicans are 32 percent, and registered unaffiliated voters are 25 percent of the total ballots cast.

However, registered Democrats are still slightly off their 2012 same-day totals, down now 2.1 percent. They have been slowly eating away at that deficit.

Meanwhile, registered Republicans are up 8.5 percent over their same-day absentee total ballots, and registered unaffiliated voters are 39.5 percent ahead of their same-day absentee total ballots from 2012.

In breakdowns by race:

White voters continue to be 72 percent of all absentee ballots in North Carolina, with black voters at 22 percent. White voters are ahead of their 2012 same day totals by 16 percent, while black voters are down 13 percent, and all other races are up 44 percent.

Women are 56 percent of the total ballots cast, with 47 percent of them registered Democrats, 30 percent registered Republican, and 23 percent registered unaffiliated. Among male voters, 37 percent are registered Democrats, 35 percent registered Republicans, and 28 percent registered unaffiliated.

The total accepted ballots now stands over 2 million:

Turning to the accepted in-person absentee ballots, nearly 2 million have come in as votes for the November 8 general election. This represents a 13 percent increase over the same day in 2012's early voting in-person period. 

Among the registered parties:

Registered Democrats are 44 percent of the accepted in-person absentee ballots, with 31 percent from registered Republicans, and 25 percent from registered unaffiliated voters. Registered Democrats are two percent down from their same-day 2012 totals, while registered Republicans are ahead by 17 percent and registered unaffiliated voters are 43 percent ahead of their same-day totals from four years ago.

The daily comparisons to 2012 are:

Among accepted in-person absentee ballots by race and party registration:

Finally, of the 2016 accepted in-person absentee ballot voters, their voting registration and voting methods from 2012 are:

More analysis to be posted later this afternoon.