The African American vote after Obama

One major part of the 2016 election that I have not addressed until now is the black vote. I wondered to what extent, if at all, African American voting behavior would change after President Obama, the inspirational first black president, would no longer be on the ballot. Republican presidential nominee Donald Trump made campaign overtures to the black community, promising to address their economic woes, asking pointedly what they had to lose.

Cutting to the chase, black turnout plummeted, but the African Americans who did vote supported Democrat Hillary Clinton almost as overwhelmingly as they had Obama. But the details below are worth reading.

Since the 1930s (1940s in St. Louis) African Americans have been strong supporters of the Democratic Party, following years of loyalty to Republicans as the party of Abraham Lincoln. The appeal of New Deal programs attracted the first wave of party switchers, and the association of Democrats (like President Kennedy) with the civil rights movement brought black support for Democrats up to the 90% level in 1964, where it has largely remained ever since. Obama's candidacy brought the double surge of nearly 99% black support and higher turnout of African American voters. The high level of support Secretary Clinton received from black voters against Vermont Sen. Bernie Sanders in the presidential primaries and President Obama's working endorsement of her candidacy in the general election foreshadowed that Clinton would continue the trend. Exit polls showed her beating Trump 88% to 8%, but when the actual votes were tabulated, results were mixed.

My analysis focuses mostly on seven St. Louis wards (1, 2, 3, 4, 21, 22 and 27) in which almost all voters are African American. Four other wards (5, 18, 19 and 26) have African American majorities but with substantial white minorities. All 11 of those wards elect African American aldermen. In St. Louis County, six townships have African American majorities, but all have larger white minorities than the city wards mentioned above. (Norwood Township, south of Ferguson, approximates the racial mix in the 18th Ward (the northern edge of the city's Central West End, north to Page Avenue). The integrated wards and townships provide less reliable evidence of black voting behavior, even though the whites there vote heavily Democratic, because 80% is a really high Democratic percentage for white voters, while black voters generally top 95%; so a white population as low as 25% still produces significant dilution of the black vote. On the other hand, black voters in the segregated wards may well vote more Democratic than black voters in integrated suburban and rural locations. But the segregated wards are the most accurate election returns available for this analysis.

Race-based comparisons to years before 2004 are complicated by significant ward and township redistricting after the 2000 census, when the segregated black north-side 20th Ward and the black majority Halls Ferry Township were eliminated and absorbed by neighboring wards and townships.

So, what happened in 2016? It's a two-part answer, one part favoring each party. The good news for Democrats is that Secretary Clinton trounced Trump in the segregated black wards, 96% to 2%. The 8% Trump support in the exit polls did not surface in St. Louis. While some black Republicans who backed Obama returned to their party, Clinton's vote fell about halfway between Obama's and John Kerry's in the election before Obama (2004). Her support was diluted down to just below 90% in the integrated black majority wards (about the same as Kerry's 2004 support) and down to 85% in the black county townships.

Changes in Democratic presidential vote share in black neighborhoods

	Kerry	Obama	Obama	Clinton
Segregated black wards	93.30%	98.39%	98.50%	95.61%
Black majority wards	89.46%	93.14%	91.47%	89.83%
Black majority townships	*	93.30%	87.36%	84.90%

Changes in Republican presidential vote share in black neighborhoods 

	Bush	McCain	Romney	Trump
Segregated black wards	3.31%	1.35%	1.32%	2.14%
Black majority wards	7.46%	6.43%	7.68%	6.08%
Black majority townships	*	10.76%	11.29%	10.84%

*Township-level election returns for 2004 in St. Louis County are not available online.


Republicans can take solace in the second part of the answer. The Obama-inspired surge in African American turnout evaporated in 2016. Compared to 2012, turnout was almost 26% lower in the segregated black wards, nearly 20% lower in the black majority wards, and nearly 18% lower in St. Louis County's black townships. Turnout in the city wards in 2016 was actually the worst of this century, dipping well below pre-Obama levels. In contrast, turnout in white majority wards and townships was mostly either level or higher, and turnout was way up in the rural and exurban areas where Trump soared. The reduction in black turnout was a major reason why Missouri, a national political barometer before 2008, was not close this time.

Changes in voter turnout in St. Louis black neighborhoods

	2004	2008	2012	2016
Segregated black wards	35563	37794	36072	26814
Black majority wards	19588	21721	18725	15070
Black majority townships	n/a	108356	106673	87533

The charts above demonstrate the consistent patterns based on how diluted the black vote is or isn't. However, some of the variations are also instructive. In the city, in spite of lower overall black turnout, Trump drew more black votes than Romney (albeit just 98 votes, a gain of less than a percentage point); but in the diluted wards and townships Trump and Clinton both lost voters (Clinton losing more in the black county townships and both losing voters in the same proportions in the integrated black majority wards). Obama's 2008 support held fast in 2012 in the segregated black wards, but his support slipped that year in the integrated wards and townships like he did in white areas.

There were also differences in how Clinton and Trump performed in relation to the rest of their party tickets. In all city wards where black voters were significant factors, Clinton led the Democrat ticket and Trump fared worse than the entire Republican ticket. But in black townships in St. Louis County, the presidential candidates were the median contests (or close to it); the county townships where Clinton led her ticket and Trump trailed his were the mostly Republican elite areas where Trump ran poorly (as discussed in this earlier post).

Without expensive and extensive survey research, the cause of these differences is subject to speculation. One possibility is that black voters living outside the echo chamber and peer pressure of a segregated ward may be more independent (or subject to different class influences or peer pressure from white neighbors). Spanish Lake Township, home to a largely black middle and upper-middle class, votes less Democratic than the other black townships and, like upper middle class white areas, was also less receptive to Trump.

Differences in the respective white minorities may also account for the variations. Black settlement and white flight in the black-majority areas of the city were fairly complete by the 1990s. The white minorities in the those areas consist primarily of (a) liberal Central West End and Midtown residents, whose neighborhoods remain predominantly white, and were redistricted into black wards to maintain the number of black aldermen, and (b) predominantly young, progressive singles and childless couples who are re-gentrifying parts of the city. Both are strong demographics for Clinton. But the black majority county townships are mostly formerly white areas where integration is still proceeding, and the remaining whites are primarily older blue-collar voters who were a promising demographic for Trump.

How Clinton and Trump affected their tickets

Heading into the 2016 presidential election, except for pockets of enthusiasm for either the billionaire populist or the potential first woman president, it was clear that both major-party candidates were really unpopular. It was uncertain, though, how down-ballot candidates would be affected. Would Republicans who couldn't stand Trump and Democrats who couldn't stand Clinton stay home and vote for no one, would they vote but split their tickets for another party's presidential candidate, or would they be so repelled by a party who would nominate such a person that they voted for most or all of the other party's candidates? We now have most of the answers. While this post examines data from just Missouri, it probably tracks similar locales in other states.

The really bad news for Democrats is the extent to which disaffected Democrat voters stayed home. As much as progressive commentators complain about disloyal Democrats voting instead for Libertarian Gary Johnson, Green Jill Stein or a write-in candidate, those voters were at the polls to vote in down-ballot contests. The stay-home voter voted for no one. Both St. Louis and Kansas City logged about 10% fewer total votes in 2016 than in 2012. The total suburban vote in St. Louis and Jackson Counties also declined. These are Democrats' strongest areas. Elsewhere else in the state, which is now mostly Republican, turnout was up.

What about voters of both parties who did show up but didn't vote for their party's presidential candidate? When comparing the presidential vote to that for the rest of the ticket, one needs to be mindful that a big gap means more ticket splitting and less benefit to the rest of the ticket (or less harm when the presidential candidate is tanking). Trump made little effort on behalf of the rest of ticket, much of whom wanted nothing to do with him. Nevertheless, most of the Trump surge also seemed to help the entire Republican ticket. Statewide, Trump led the state-office ticket by just a little over a point, and Republicans swept all statewide offices for the first time since 1928.

In the last big Republican win, Ronald Reagan won 60% of the Missouri vote in his 1984 re-election, a nearly 9-point improvement over 1980. The Republican state-office ticket won 52.7% in 1984, a 6+ point improvement over the prior election. But Reagan led the rest of the Missouri Republican ticket by more than 7 points, and Democrats retained the open Lieutenant Governor's seat (and control of both houses of the state legislature). So, in spite of everything, Trump appears to have helped his Republican ticket mates more than Reagan helped his.

It would be a mistake, however, to credit Trump (or blame Clinton) entirely for Republicans' ticket-wide success. While Trump ran 3 points ahead of Romney statewide, Republican candidates for the five statewide offices on the ballot (excluding the U.S. Senate race) surged an average 10 points ahead of their 2012 counterparts. These down-ballot results are partially explained by the respective situations of the down-ballot contests and the candidates who ran. In 2012, popular incumbents (3 Democrat, 1 Republican) ran in four of the five contests, but in 2016 all five contests were open seats. Nevertheless, the magnitude of the 10-point statewide surge was probably somewhat influenced by the presidential vote.

The statewide comparisons, though, are necessarily averages of the entire state, where one area's movements in one direction are canceled out by another area's movement in the opposite direction. (I wrote earlier about pockets of Trump's appeal to blue-collar Democrats and his repellent to upper-crust establishment Republicans.) These areas produced a somewhat greater coattail effect. In Trump's strong areas, the whole Republican ticket surged over 4 years ago, but Trump still ran several points ahead of the ticket (but still less than Reagan in 1984), leaving some potential Republican votes on the table. In Clinton's pockets of strength in urban and suburban areas (and Boone County, home of the University of Missouri and the only rural county Clinton carried), Trump declined compared to Romney, and the statewide ticket's improvement was much less than in the rest of the state.


Details: In the rural and exurban counties, where Trump led the Republican ticket, the statewide ticket improved by nearly 13 points. But those areas also experienced more ticket splitting, as Trump led the ticket by 5½ points in rural counties and 4 points in exurban counties. The three white St. Louis County Townships where Trump improved over Romney also saw the state ticket improve by 5.8 - 9 points, which was better than in the rest of the county, and there wasn't much ticket splitting. Results were mixed in the city neighborhoods where Trump improved. The more Democratic areas had high ticket splitting and the state ticket improved in line with the rest of the city, but the state GOP ticket surged in "city limits precincts" in Wards 12, 23 and 24. In other urban and suburban areas and Boone County, where Trump trailed the rest of the Republican ticket by about 2½ points, the state candidates' improvement (4 points urban, 6 points in Boone County and 6½ points in the suburbs) was less than in areas where Trump did well. In the St. Louis County townships were Trump underperformed Romney the most, the statewide ticket's performance also improved the least (1.1% in Clayton Township and 1.6% in Hadley Township).

Isolated areas where the state Democratic ticket bucked the trend and improved over 2012 probably represent population changes such as white flight more than political trends. This was notable in downtown St. Louis and the Loft district, which have both experienced well-publicized increases in crime.

The big question has yet to play out. Will the Trump Democrats and NeverTrump Republicans undergo a permanent party change, or was 2016 just a one-time thing? The success or failure of the Trump presidency will have a lot to do with the answer.

Diverging cross-currents in the 2016 election

I wrote earlier how Trump's win in Missouri, as well as nationally, was fueled by a surge in support from rural and exurban areas, but that this surge merely continued the direction set in earlier 21^st Century presidential elections. However, this continuity masks the underlying intraparty strife that played out in 2016. Even the third-party challenges of Ralph Nader and Pat Buchanan in 2000 didn't touch the partisan uncertainty that prevailed in 2016.

One tea party blogger attributed the Trump surge to the common man's electoral revolt against the elites. Maybe so, but that revolt was just a reaction to the elites' own overwhelming rejection of Trump. We would have read all about it as the cause of Trump's defeat, if Trump had lost as expected. Trump's unexpected triumph changed the focus to how such a thing could have happened. But I will examine the elite revolt that other commentators have forgotten.

This is not about how millennial Democrats and Democrat elites in media, academic and old-money circles hated Trump even worse than prior Republican nominees; their votes for Democratic candidates remained consistent, and the intensity of their disdain for Trump didn't make their votes count any more. What drove Trump's performance in certain areas below Mitt Romney's four years earlier was Republicans and Republican-leaning independents changing how they voted.

This post analyzes data from the City and County of St. Louis because those are the areas where I know enough about the neighborhoods producing the results to make meaningful conclusions. Similar trends probably occurred in other urban and suburban areas nationwide, so these observations may be useful on a national level.

Trump's rejection by his party, predictably, was greatest in St. Louis' central corridor, both city and county. As I noted in an earlier post, the most dramatic shift took place in Clayton Township (mostly western Clayton and Ladue), where a half-point Romney win turned into a 19-point Trump loss. It was one of only two townships were Secretary Clinton picked up more of the lost Republican votes than third parties and write-ins did. Trump's next biggest drop came in Missouri River Township (Town & Country), where Trump lost 12 points compared to Romney (but still won). In progressive, formerly Republican Jefferson Township (Webster Groves), Trump suffered an 11-point decline, but most of it went third-party. Another predictable area of Trump decline was Creve Coeur Township, where he dropped nearly 9 points. Trump's greatest decline in the City of St. Louis was the 28^th Ward (the very old-money Central West End and Skinker-DeBaliviere), where he dropped nearly 7 points. (Declines in the City are less dramatic because the electorate there is already so Democratic as to leave little room for movement towards Democrats.) In all five of those jurisdictions, Trump trailed the entire statewide Republican ticket (and Clinton led the Democrats).

While Trump's rejection by moderate pro-business and old-money Republicans was expected, the breadth of Republican flight from their populist nominee was surprising to me. Trump also lost a lot of ground in “new money” west county. Trump dropped 10 points in Chesterfield Township, nearly 9 points in Lafayette Township, and 7½ points in both Wild Horse and Maryland Heights Townships. Trump ran last on the ticket in Chesterfield and below median in the others, but still carried all but Maryland Heights.

Trump also suffered lesser Republican flight in areas not usually associated with “political correctness,” such as south county and conservative wards in the southwest part of the city. Clinton picked up almost 2 points of the 6½ points Trump dropped in the City's 16^th Ward (St. Louis Hills and western Southampton). Trump dropped nearly 7 points in both Bonhomme (Kirkwood) and Gravois (parts of Affton and Crestwood) Townships, but most of those votes went third-party instead of to Clinton. In most south city wards, as well as Republican Tesson Ferry and Oakville Townships, both Clinton and Trump lost share to third-party candidates. While Trump's performance was median or lower in these areas, he did not trail the entire ticket in any of them.

Democrats suffered from their own intraparty defections. Secretary Clinton suffered more from a decline in voter turnout in many Democratic wards and townships than from actual defection to Trump. Turnout losses may be attributable to millennial voters and former supporters of Sen. Bernie Sanders who Secretary Clinton failed to win over, as well as African Americans no longer inspired without the first black president on the ballot.

Though less significant, the well-publicized defections of blue-collar whites to Trump did occur in a few select neighborhoods. Clinton's big wins in Democratic areas camouflaged Trump's gains. At first blush, the 11th Ward (where Clinton beat Trump 67%-27%), 25th Ward (79%-16%), Midland Township (60%-33%) and Airport Township (65%-29%) do not look discouraging for Democrats or hopeful for Trump. But all of these results represent a 6-7½ point drop for Clinton from Obama and a 2-3 point gain for Trump over Romney. Lemay Township (adjacent to the City's 11th Ward) sported the biggest Republican gain, with Trump gaining 3½ points and Clinton dropping 7½ points, enough to give Trump a 2-point win (48%-46%) in the usually Democratic township.

Sample trends in St. Louis County

	2012 Obama	Romney		2016 Clinton	Trump
Clayton Township	48.84%	49.47%		55.70%	36.51%
Lemay Township	53.16%	44.14%		45.67%	47.51%

Precinct-level returns disclosed some pattern differences within wards. In St. Louis' 11^th Ward, Trump advanced most in the Patch neighborhood across the city limits from Lemay and a neighboring precinct in Carondelet, but Clinton advanced in the precinct that includes upscale parts of Holly Hills. In the neighboring 12^th Ward, Trump improved in the two precincts bordering the city limits. One of them (southwest of Saints Peter and Paul Cemetery between Gravois Avenue and Morganford Road) was the City's only precinct where Trump won.

Some precincts of relative Trump strength form another interesting pattern. Trump improved (and brought much of the rest of the Republican ticket with him) in most precincts bordering the city limits, from Ellendale (bordering Maplewood) south to the Mississippi River. In addition to the 11^th Ward Patch precincts and the 12^th Ward precincts mentioned above, Trump also showed improvement in the Lindenwood Park precinct in the 23^rd Ward that includes the Shrewsbury Metrolink station and the usually progressive Ellendale precinct in the 24^th Ward bordering Maplewood. These precincts in the 12^th, 23^rd and 24^th Wards were the only precincts in those wards (except for statistical noise in one tiny precinct) in which Trump ran better than Romney. All of those precincts also produced the greatest Republican improvement in their wards over 2012 for the rest of the statewide ticket. None of these border precincts are upscale neighborhoods, and they were not among the south side's stronger Republican precincts until recently.

Whether these movements represent permanent party shifts or were merely reactions to unpopular candidates remains to be seen.