Missouri's urban-rural split on steroids

The urban/rural electoral split in Missouri got even wider in 2016.

I have written twice before about the widening urban-rural electoral split in Missouri. In 2004 my focus was putting the evangelical vote in historical context, but the upshot of that post was the contrast of the Democratic trend in St. Louis and Kansas City and their suburbs with the Republican trend in exurban and rural counties. The diverging directions of those areas resulted in a widening difference in their vote. I updated the post in 2012, noting that the intervening two presidential elections had continued the trend and widened the difference.

So, what would happen in 2016? Would Hillary Clinton's appeal (and Donald Trumps' repellent) to women everywhere and Trump's appeal to blue-collar voters put an end to that trend? In a word, no. In fact, Trump's appeal in exurban and rural areas not only continued the divergence, but sharply accelerated it. As I inferred in my post-election posts earlier this month, more of the undegreed blue-collar Democrats and Democratic leaning independents who defected to Trump live in rural and exurban areas than in urban and suburban areas.

Democrats performing better in urban/suburban areas than in exurban/rural areas in Missouri has been true for about a hundred years, but not by all that much. As recently as 1992, when Bill Clinton dispatched President George H.W. Bush, Clinton's performance in urban and suburban areas was just 9 points better than in rural areas. My prior posts linked above noted that this urban/rural split increased to 12 points in 1996, to 16 points in 2000, 21 in 2004, 23 in 2008 and 25 points in 2012. This year the split skyrocketed to 33 points. While the Democratic share of the vote dropped 2-1/2 points from 2012 in urban and suburban areas, it tanked by almost 9 points in rural counties.

The rural vote was a Tale of Two Clintons. In 1996, with many votes being diverted to third-party populist Ross Perot, President Bill Clinton won a healthy plurality in Missouri's rural counties, winning about 43% in those counties. Twenty years later his wife would win less than 25% there.

Trump's curious appeal to rural voters this year was foreshadowed by his rural success in the presidential primaries. But the continuation of the generation-long trend suggests that fundamental shifts in party identification are taking place.

The urban/rural split I have identified may actually be understated, because my county classifications have become outdated. My original 2004 post examined data starting in 1952. (See the original 2004 post to see data back to 1952.) For consistency purposes, my analysis continued to categorize counties the same way throughout the following 64-year period. But population changes since then have rendered some of my classifications obsolete. Some of the areas that I considered "exurban" then (e.g., St. Charles County outside St. Louis and Clay and Platte Counties north of Kansas City) are now behaving more like suburbs, and some counties that I considered "rural" (e.g., Lincoln and Warren Counties outside St. Louis) are now part of the St. Louis Standard Metropolitan Statistical Area (SMSA), with voting patterns that more closely resemble exurban counties than rural counties. Today, suburbs are more Democratic than exurban counties, which in turn are more Democratic than virtually all rural counties.

Moreover, the biggest rural county, Greene, has evolved into a combination of urban (Springfield, population 160,000, though just 4.1% African American and 3.7% Latino), suburban, college town (Missouri State and Drury) and exurban areas. Collectively those areas cause Greene County to be 12 points more Democratic than purely rural counties. In addition, eight rural counties containing college campuses run several points more Democratic than other rural counties. Reassigning all of these counties would make all three categories less Democratic, but the rural counties more than the others.

At the other end of the spectrum, the "urban" category has included the cities of St. Louis and Kansas City together with their inner-ring suburbs in St. Louis and Jackson Counties. I didn't segregate the city vote from the suburbs because election data before 1996 reported all of Jackson County (both Kansas City and its suburbs) together as one. Enough elections have now taken place since 1996 to provide a meaningful pattern and a more accurate comparison of rural with purely urban areas.

I have more number crunching to do, but these refinements are bound to produce even higher Democratic numbers in purely urban areas and even higher Republican numbers in purely rural counties, resulting in an even larger urban/rural split. More to come.

Analysis of 2016 presidential election in St. Louis

To most of us who live in St. Louis, Democrat Hillary Clinton seemed to be cruising to an easy win. Almost everyone we knew seemed to be for her, some enthusiastically and others readily settling for her as the obvious antidote to Republican Donald Trump. Our Republican friends couldn't stand Trump and were defecting.

And then, WHAM! Trump pierced the "blue wall" and rode Pennsylvania, Wisconsin and Michigan to an Electoral College landslide. Secretary Clinton's 2-point win in the meaningless popular vote was a mere consolation prize. Here in Missouri, despite how it seemed here, Trump won the state by 19 points and carried the entire statewide Republican ticket (including embattled Sen. Roy Blunt) in with him.

As was the case in 2004, when Democrat John Kerry seemed assured of making George W. Bush a one-term president like his father, we in St. Louis suffered from myopia. We had no clue what was going on out in the sticks. It turned out that, even more than in 2004 (or any other year for that matter), rural and exurban Missouri went big time for the Republican nominee.

An examination of results in St. Louis wards and townships shows an entirely different story. (I expect a similar analysis of Kansas City wards and Jackson, Clay and Platte County townships would show something similar.) In metropolitan areas in Missouri, there were clear patterns of both  blue-collar Democrats defecting to Trump and highly educated Republicans abandoning him. Here under "The Bubble," NeverTrumpers won that battle.

With national and state exit polls reporting that 8% of blacks (and 13% of black men) and 52% (up 5 points from 2012) nationally (and 59% in Missouri) of voters without a college degree voting for Trump, I thought he might do much better here (especially with the first black president no longer being on the ballot). But the exit poll results were not reflected in election returns here. Trump won only 2% of the vote in segregated black wards. (Returns from other black majority wards and all black majority townships are too diluted with blue-collar whites to be a meaningful measure.) All told, Secretary Clinton fared nearly as well here as President Obama had four years ago. Both Clinton and Trump lost ground here compared to Obama and Romney. Third-party candidates and write-ins picked up the slack. Clinton dropped more in the city and Trump dropped more in the county.

Patterns of both Trump and NeverTrump strength emerged in the ward-by-ward and township-by-township data, especially when compared to past presidential elections.

Where NeverTrumpers powered Clinton gains

First, let's examine Secretary Clinton's areas of relative strength compared to Obama. On the whole, in the St. Louis area, defections of NeverTrump Republicans overwhelmed the counter-movement of blue-color Democrats and Democrat-leaning independents voting for Trump. As accurately predicted by polls, these were largely areas with more highly educated voters, a national demographic that favored Clinton. Most were in Republican townships where Trump's drop was larger than Clinton's gain. Her biggest comparative success was in Clayton Township (western Clayton, Ladue and Rock Hill), formerly the county's most Republican township. It followed the national Republican decline in old-money trust-fund neighborhoods (but about a generation behind the east and west coasts) so that by the turn of the 21st Century, few Republican candidates carried the township. Sen. John McCain won 43.7% in 2008 and Mitt Romney claimed a narrow plurality win in the township with 49.7% in 2012. But Trump's numbers plummeted 13 points to 36.51% while Clinton picked up 7 points over Obama.

Clinton gains and Trump drops aren't evident at first in Republican areas because Trump still carried many of them. But when compared to 2012, in addition to Clayton Township, Clinton scored big gains and Trump suffered significant deterioration in Missouri River Township (Town & Country), where Clinton gained 6 points and Trump dropped 12, followed by Jefferson Township (Webster Groves) (Clinton up 5 points, Trump down 11), Chesterfield Township (Clinton up 5 points, Trump down 10), Creve Coeur and Lafayette Townships (Clinton up 3 1/2 points and Trump down 9 in both), Hadley Township (eastern portions of Clayton, Richmond Heights and University City) (Clinton up 3 points and Trump down 8 1/2), Maryland Heights and Wild Horse Townships (Clinton up 2 1/2 points and Trump down 7 1/2 in both), the City's 28th Ward (Central West End) (Clinton up 2 points and Trump down 7), Ward 16 (St. Louis Hills) (Clinton up 2 points and Trump down 6 1/2) and Ward 19 (Grand Center and St. Louis University) (Clinton up 2 1/2 points and Trump down 5 1/2). In all but three of those jurisdictions, as well as in all of the city's African American wards, Trump ran worse than every other statewide Republican on the ballot. If you live in any of these areas, your expectations of a Clinton win were based on an accurate perception of what was really happening in your area.

Blue-collar gains for Trump

What about white voters without degrees? In a mirror image of what happened in Republican parts of St. Louis County, Trump gains and Clinton drops aren't evident at first in blue-collar areas because Clinton still carried them. Nevertheless, blue-collar whites are probably the reason Trump improved over Romney's numbers (and Clinton underperformed Obama) in Wards 9 (Benton Park and Mount Pleasant neighborhoods), 10 (The Hill), 11 (Carondelet and the Patch), 13 (Holly Hills), 20 (Marine Villa and the area south of Cherokee Street) and 25 (Dutchtown), all in south St. Louis. Trump showed similar improvement in Airport (St. Ann), Midland (Overland) and Lemay (Lemay and Mehlville) Townships in St. Louis County. The most significant were the 11th Ward and Lemay Township, both of which provided Trump with a 3-point gain and Clinton with a 7-point drop. The improvement was marginal (about half the 5-point increase in the exit polls), but was notable because it ran counter to the trend in the city and county. Those results were also probably diluted by NeverTrumpers in those neighborhoods moving in the other direction. Trump didn't trail the entire Republican ticket in any of these Trump-improvement wards or townships, running second-best in the 11th Ward and (except for the racially diverse 9th and 20th Wards) no worse than median in the others. As was the case in the city's 14th Ward (Bevo), Trump's improvement in south county was probably negated or diluted by a significant Bosnian Muslim population that I think feels threatened by Trump's policies on national security. However, the significant Latino presence in the 20th Ward did not retard Trump's improvement there (consistent with exit polls amazingly showing Trump improving slightly over Romney's performance with Latinos).

In addition, Trump's poor showing in black wards was actually better than either Romney or McCain had managed against Obama the prior two elections. Secretary Clinton dropped nearly 3 points there, but Trump was only able to pick up 3/4 of a point. The impact of the black vote going all the way up to 2% was minor. More significant was the 29% drop in turnout in the segregated African American wards, reducing Clinton's margin there by nearly 10,000 votes.

Outstate voters powered Trump's big win

How did Trump win the state? Outstate voters love him. Apparently most of the blue collar voters who broke for Trump live in Missouri's rural counties and the exurban "collar counties" surrounding St. Louis and Kansas City. No significant erosion of Trump's base was evident in those results, except in counties containing either a major city (like Springfield) or a college campus (like Mizzou or Truman State). In the exurban counties, Trump won 67% of the vote. Even Jefferson County south of St. Louis, long a Democrat stronghold, went 65% for Trump. Rural voters love Trump even more, as Trump outperformed the rest of the statewide Republican ticket in most rural counties. Trump won 75% of  voters in rural counties (excluding cosmopolitan Greene County and rural counties with college campuses). He topped 80% in 20 rural counties, including two (Mercer (a "Dukakis county") in northern Missouri and Bollinger west of Cape Girardeau) where he topped 85%. That's more dominant than Democrats are in the City of St. Louis! All of these percentages dwarf the numbers of prior Republican presidential nominees.I am planning a later post that examines the county-by-county returns more thoroughly and puts them in historical context, but I still have numbers to crunch.

In short, the exurban and rural Missouri surge for Trump overwhelmed NeverTrump trends in metropolitan St. Louis and Kansas City.

'Dukakis counties' illustrate Democrats' decline in rural Missouri

Donald Trump's upset win in the 2016  presidential race was based in rural counties in "flyover country," places that the "deplorables" call home. The national red-blue county map looked much the same as prior 21st Century presidential elections, but Trump's rural margin was dramatically higher.  Most of the under-educated blue-collar whites who powered Trump's win lived in rural and exurban areas, while similar voters in urban and suburban areas mostly stayed with Democrat Hillary Clinton. Secretary Clinton won big in St. Louis City and County and in Kansas City. But she lost every rural or exurban county except Boone, home to the large (and progressive) academic community at the University of Missouri.

What happened? Democrats formerly held their own quite well in rural Missouri. The state re-fought the Civil War at the ballot box every four years, with counties that had wanted to join the Confederacy (especially in the southeast Missouri bootheel and the Little Dixie region in northeast and central Missouri) voting Democrat and Union-loyal counties in southwest Missouri and the German counties along the eastern Missouri River voting Republican. This pattern mostly lasted through the end of the 20th Century.

Republican gains and setbacks in rural Missouri alternated throughout the Eisenhower, Kennedy and Johnson administrations. The foundation for a permanent shift was laid in 1968, when Richard Nixon beat Hubert Humphrey and George Wallace by winning the suburbs and some traditionally Democratic rural counties, with one bootheel county (Stoddard) defecting to Nixon and another (Pemiscot) going full rogue for Wallace. In Nixon's 1972 landslide reelection over anti-war Democrat George McGovern (who first picked then dumped Missouri Sen. Tom Eagleton as his running mate), Nixon won every rural country except Monroe. Those fortunes reversed abruptly but temporarily in 1976, when Watergate reaction and the candidacy of born-again Christian Democrat Jimmy Carter brought Democrat numbers in rural counties to their modern-day high. Carter's perceived betrayal of conservative Christians sent them fleeing to Ronald Reagan in 1980, giving them a new political home that has persisted to this day. That brings us to 1988, when George H.W. Bush won most rural counties, leaving the 33 rural Missouri counties who voted for the inept Michael Dukakis as the last hard-core "blue dog" Democrat holdouts that I examine in this post.

I chose 1988 as my base line because in the next two elections, populist billionaire independent (later Reform Party) candidate Ross Perot put a major dent in both parties' vote totals, murking the two-party trend lines. Democrat Bill Clinton won Missouri both times, including several rural Missouri counties, but mostly by mere pluralities, as Perot's votes came more from erstwhile Republicans than Democrats. During those eight years of presidential statistical noise, many fundamental pro-Republican changes occurred. Just two years in, Republicans won control of Congress for the first time in a generation. Later, Clinton sex scandals and the partisan divide over his impeachment would make rural Christian voters even more Republican. While Perot had effectively throttled the elder Bush's re-election and handed the White House to Bill Clinton, Perot also served as the bridge to Republican dominance thereafter. In 2000, when Perot declined to run again, most of his Republican supporters returned to the GOP fold, but many of his Democrat supporters either stopped voting or crossed over to the Republicans. That year, GOP presidential candidates began a rural-based winning streak in Missouri that persists to this day.

Returning to 1988 as my baseline, I compared the 33 rural counties Dukakis carried in 1988 with their numbers this year, and the comparison is jaw dropping. Dukakis won Mercer County on the Iowa border, but by 2016, Hillary Clinton won only 12.38% of that county's vote. And the 37-point drop there wasn't even the state's largest. About 400 miles south in the lead belt, Reynolds County dropped nearly 44 points, giving Dukakis 61.42% but only 17.81% to Secretary Clinton. In Monroe County in northeast Missouri's "Little Dixie," the sole rural Democratic holdout in 1972, Democratic presidential performance declined steadily from 1988 to 2016. Dukakis' solid 61.31% there dwindled nearly 41 points over the next 28 years to Hillary Clinton's 20.53%. Neighboring Ralls County dropped just as much.

The median 1988-to-2016 decline among the rural "Dukakis counties" was 29.4%. Here are the ten largest declines in Democrat fortunes among those counties:

County Dukakis 1988 Clinton 2016 Democrat decline Reynolds 61.42% 17.81% 43.61% Monroe 61.31% 20.53% 40.78% Ralls 62.38% 21.61% 40.77% Mercer 50.00% 12.38% 37.62% Lewis 57.57% 20.96% 36.61% Oregon 54.22% 18.65% 35.68% Shelby 53.35% 18.62% 34.73% Dunklin 54.53% 19.93% 34.60% Clark 56.11% 21.83% 34.27% DeKalb 51.26% 17.94% 33.33%

Republican rural counties got even more Republican over that period, but the change was less dramatic. In Wright and Douglas Counties in southwest Missouri, Secretary Clinton's declines from 1988 were only about 20 points. In Jasper (Joplin) and Gasconade (Hermann) Counties, her declines were even smaller.

A major counter trend in urban and suburban areas has kept Democrats competitive in Missouri. Hillary Clinton carried St. Louis County this year, 55% to 39%, a margin of over 81,000 votes. In 1988, Bush had carried St. Louis County by nearly 10 points, a margin of over 46,000 votes. Bush's big suburban win then wasn't unusual, as the Republican presidential nominee had won St. Louis County every prior election since Lyndon Johnson's 1964 drubbing of Barry Goldwater. But no Republican presidential nominee has carried St. Louis County since Bush's win in 1988.

The core cities of St. Louis and Kansas City have also moved even more Democratic, but their impact is blunted by their shrinking populations. While Secretary Clinton's 79% in the City of St. Louis was six points better than Dukakis, that only improved her victory margin over 1988 by 12,000 votes. With nearly 2.8 million votes being cast in 2016, her improved St. Louis performance improved her statewide share by less than half a percent.

The widening gap between rural voters and urban and suburban voters in Missouri has been in process since 1976, but it accelerated in 2012 and 2016. I will have a more thorough analysis, hopefully with a graph, in a later post. Stay tuned.