KEY POINTS FROM THIS ARTICLE

— In 2016, Donald Trump inspired higher Republican turnout in Pennsylvania, while Hillary Clinton couldn’t offset her losses in the non-metro parts of the state.

— Voter registration trends in Pennsylvania are mirroring the 2016 picture — all of the counties in Philadelphia’s suburban collar are Democratic by registration while Republicans have flipped some working class counties.

— With the third party vote projected to be down from 2020, former Gov. Bill Weld’s (R-MA) relative strength as a Republican protest presidential candidate in this month’s Pennsylvania Republican primary may be a warning sign for Trump.

— Joe Biden, who frequently talks up his working class Scranton background, gives Democrats a good chance to move the state back into the blue column, but it’ll hardly be an automatic shift.

States of Play: Pennsylvania

The drive from Hagerstown, Md., north to State College, Pa., is about 110 miles. Two-lane roads connect the two, running through bucolic farmland, over forested mountains, and in and out of a succession of small towns, with names such as Burnt Cabins, Shade Gap, Shirleysburg, and Orbisonia. There are occasional signs of prosperity, but many of these communities are economically strapped. It is “God and guns” country. And in 2016, it was also Donald Trump country.

Early that October, this author made the drive, counting yard signs along the way. And upon reaching State College, the tally stood at two signs for Hillary Rodham Clinton and 68 for Trump. To be sure, the drive was basically through a Republican swath of Pennsylvania. Yet the number of Trump signs was unusually high for a presidential election, and were up more than a month before the November balloting. They did not appear to be the work of one person, or even a small team of Trump campaign workers, but of individuals of like mind expressing a similar sentiment.

Counting yard signs is not scientific. It is not a poll. But the result of this personal count that day spoke to an intensity of support for Trump’s anti-establishment, outsider candidacy in this slice of rural, small-town America. The numerous signs gave the sense that something big was happening, but what that might be was unclear until Election Day, Nov. 8, 2016.

From 1992 through 2012, Democrats had made Pennsylvania a critical part of their Rust Belt “firewall,” winning it in six straight presidential elections. They had done so by executing a simple formula: Build a large lead in Philadelphia (with its sizable minority population), add to the margin in its adjacent suburbs and across the state in Allegheny County (Pittsburgh). Then, fend off a Republican counter-surge in the more conservative remainder of Pennsylvania with its smaller cities, old industrial centers, and rural terrain.

As Election Night 2016 began to unfold, it looked as though the Democratic formula would work once again. Clinton rolled out of Philadelphia with a huge lead of 475,000 votes, added another 188,000 votes to her margin in its suburbs, and then an additional 108,000 votes in Allegheny County. Taken together, these large Democratic metro centers had more than done their part, giving Clinton a whopping lead of 771,767 votes. It far surpassed Barack Obama’s 706,314-vote advantage in these same places in 2012, en route to an Obama statewide victory of more than 300,000 votes.

However, as Election Night progressed, the unexpected happened. Clinton’s lead dwindled and dwindled and dwindled, until Trump finally pulled ahead and won Pennsylvania by the thin margin of 44,292 votes out of more than 6.1 million cast. The “hinterland” had outperformed the populous “cosmopolitan” centers at the ballot box.

But it took a two-way flow of votes to ensure Trump’s victory. The statewide Republican presidential vote was up dramatically from 2012, to a Pennsylvania GOP record of 2.97 million, while the Democratic presidential vote dropped to 2.93 million, the party’s lowest presidential tally in the Keystone State since 2000. And this two-way flow — the Republican vote up, the Democratic vote down — was on steroids in Pennsylvania’s vast interior.

Altogether, Trump’s vote tally was nearly 300,000 more than Republican standard-bearer Mitt Romney’s total in 2012, with all of the gains coming in Pennsylvania’s heartland. Meanwhile, in that same territory, Clinton’s total was down by more than 125,000 votes from Obama’s total four years earlier. Taken together, the Republican margin outside the Philadelphia and Pittsburgh areas more than doubled from 2012, exploding to more than 800,000 votes. It was just enough to overturn Clinton’s huge lead in the large metro centers.

Trump’s new coalition

When Republican presidential candidates were carrying Pennsylvania in the 1980s behind Ronald Reagan and George H.W. Bush, they tied together the Philadelphia suburbs, smaller urban counties such as Lancaster, Lebanon, and York, along with small towns and rural areas of the state’s interior. In 2016, with the suburbs trending Democratic, Trump had to splice together a different coalition that accented a combination of the historically WASPish Republican towns of central Pennsylvania with heavily ethnic, formerly Democratic industrial strongholds in northeast, northwest, and western Pennsylvania.

Of the state’s 67 counties, 61 are in this part of the state, of which Trump carried 56. To be sure, the Republican trend in the industrial areas had been under way for years. But Trump increased the volume of the GOP presidential vote in every one of the 61 counties from 2012. Often, it was by a large amount. In Luzerne County (Wilkes-Barre), in the coal country of northeast Pennsylvania, the number of Republican presidential ballots shot up by 35%, from less than 60,000 in 2012 to nearly 80,000 for Trump. The surge enabled him to be the first GOP presidential candidate to carry Luzerne since 1988. He also flipped Erie County, in Pennsylvania’s northwestern corner, from Democratic in 2012 to Republican in 2016, as well as Northampton (Bethlehem) in the Lehigh Valley north of Philadelphia.

Altogether, Trump drove up the number of Republican presidential votes by at least 10 percentage points from four years earlier in all but a half dozen of the 61 “interior” counties, and by at least 25 points in more than a dozen of them.

The new Trump coalition, a combination of life-long Republicans, disgruntled Democrats and independents, and non-voters, has been called angry, aggrieved, and overlooked. Trump assiduously courted them with his brash sense of confidence, promising to renegotiate what he saw as “job killing” international trade deals that would ultimately bring old and new industries alike to Pennsylvania. But he also connected with voters in the state’s heartland by expressing his affinity with their culture of church, country, and guns, and he pledged to be their voice in Washington.

The Trump campaign’s courtship of these voters in 2016 was hard to miss, from the ubiquitous campaign signs to high-decibel rallies in arenas “off the beaten path.” The effort was so successful that by Election Day that year the large Trump vote was solid and hidden in plain sight.

But even this may not have been enough to carry Pennsylvania if Democrats had not conspired in their own defeat. If Clinton had just been able to match the 2012 Obama vote — 2.99 million — she would have been able to beat back Trump in Pennsylvania. Instead, the Democrats were well along in the process of trading the votes of socially conservative blue-collar workers for more socially liberal white-collar voters in the affluent suburbs. Through six straight Democratic victories, it had been a worthwhile trade as the party held enough of the non-metro blue-collar vote each election to make it work.

Yet in 2016, Clinton was in poor position to compete with Trump for these voters. Her platform — pro-gun control, supportive of federal action on climate change, and anti-coal — was fashioned for the nation’s large cities and suburbs. And a couple of maladroit remarks in 2016 made her situation even more untenable. In a state with a strong coal mining heritage in its northeastern and western parts, Clinton inexplicably said in March 2016 that “we’re going to put a lot of coal miners and coal companies out of business.” That September, she added to the damage by declaring at a New York City fundraiser that half of Trump’s backers belonged in a “basket of deplorables.” The remark was not explicitly aimed at Pennsylvania voters outside the state’s large cities and suburbs, but it might as well have been.

All this followed by eight years a provocative comment by Obama at a San Francisco fundraiser about voters in small-town Pennsylvania who “cling to guns or religion or antipathy toward people who aren’t like them… as a way to explain their frustrations.” Given the starkness of the remark, Obama was fortunate to run as well as he did in 2008 and 2012 outside the Philadelphia and Pittsburgh areas. Yet his comment and those of Clinton underscored a sense of growing estrangement between Democrats and their once loyal blue-collar voters that may not be easy to repair in 2020.

Looking ahead to November

In Joe Biden, Democrats are probably fielding the best candidate they could this year to reach Pennsylvania’s “Trump country” voters. Throughout his political career, Biden has portrayed himself as an outgoing “average Joe,” comfortable with mingling with blue-collar workers in their ethnic halls and VFWs. For good measure, he was born in Pennsylvania, Scranton to be precise. And while his family moved to Delaware when he was a boy, his home base of Wilmington is just a few miles south of the Pennsylvania border within the Philadelphia media market. Over the years, Biden has come back often to Pennsylvania to campaign for Democratic candidates, and he has established his national campaign headquarters in Philadelphia. He will surely try to run this year as the state’s favorite son.

As of June 26, the RealClearPolitics rolling average of polls showed Biden with a lead of 6.3 percentage points over Trump in Pennsylvania. That, however, must be taken with a grain of salt. Four years ago, RealClearPolitics showed Clinton on Election Day with a lead in Pennsylvania that was down to 1.9 points, before Trump carried the state by seven-tenths (0.7) of a point. Apparently, many of the late deciders in Pennsylvania (as elsewhere) broke for the Republican. It is also quite possible that the size of the Trump vote in 2016 was routinely under polled because either a number of voters did not feel comfortable telling pollsters that they were for the brash Manhattan billionaire or were genuinely undecided until the end and broke toward the challenger. That could happen again in 2020.

During the Trump presidency, Pennsylvania Republicans have chipped away a bit at the Democrats’ wide voter registration advantage. At points during the Obama years, registered Democrats outnumbered registered Republicans by more than 1 million. The Democratic lead had dropped to around 915,000 when Trump was elected in 2016 and this June was barely 800,000. (In addition, more than 1.2 million Pennsylvanians are registered as independents or as members of third parties.)

As a statewide figure, the sizable Democratic registration advantage is rarely matched by a similar degree of Democratic success at the polls. More significant are the registration trends, with each party since Trump’s election picking up additional voters in strongholds that are fairly new to the party — for Republicans, the once Democratic industrial areas; and for Democrats, the once Republican suburbs of Philadelphia.

For the first time ever, the four suburban counties all have Democratic registration leads, with Chester the last of the four to flip to the Democrats this May. In Bucks County, the Democratic registration edge has increased from less than 10,000 in November 2016 to more than 15,000 in June 2020. In Delaware County, what was a Democratic edge of less than 20,000 when Trump was elected has swelled to more than 40,000 this month. And in Montgomery County, once the cornerstone of the Pennsylvania Republican Party, a Democratic registration margin approaching 55,000 in November 2016 has now blown past 85,000.

On the other hand, Republican registrations are surging in pockets of blue-collar strength outstate. In Beaver County, along the Ohio border, a Democratic registration lead approaching 20,000 at the time of Trump’s election is now barely half that. In Cambria County (Johnstown), a Democratic registration advantage exceeding 12,000 in November 2016 is barely 2,000 this spring. And in Westmoreland County outside Pittsburgh, the registration lead has flipped from Democratic to Republican during Trump’s three-plus years in the Oval Office.

Meanwhile, the state’s presidential primary June 2 gave Biden and Trump a chance to test their vote-getting appeal in Pennsylvania. By and large, both passed their test with flying colors in contests that were largely non-competitive. Biden drew 79% of the 1.59-million Democratic primary votes against Sen. Bernie Sanders (18%) and Rep. Tulsi Gabbard (3%), both of whom had already quit the Democratic race and endorsed Biden. Trump won 92% of the 1.14-million Republican primary votes against former Massachusetts Gov. William Weld (6%) and perennial candidate Roque “Rocky” De La Fuente (2%).

The votes for Weld could be a problem for the president come fall as they were cast by GOP voters for a candidate who was viscerally anti-Trump. Weld’s total of nearly 70,000 votes in the Republican primary in Pennsylvania represented his highest total in the entire 2020 primary season and was 25,000 votes more than Trump’s 2016 margin of victory in the state. It would be surprising if there were more than a small handful of Republicans who voted for Weld in the spring and would then turn around and vote for Trump in the fall. More likely, many of Weld’s voters would end up with Biden, if they vote for president at all.

Still, it will be a tall order for the Democrats to defeat the president in Pennsylvania. He has carried the state once already and has cultivated it since then. And he has already signaled that he will not relinquish Pennsylvania and its 20 electoral votes without a tough fight. Case in point: his appearance at a town hall this March in Biden’s home town of Scranton. There, the 74-year-old Trump launched a sharp volley at the 77-year-old Biden, questioning his mental competence to be president.

Meanwhile, since the coronavirus has taken hold in the United States, Biden has made several trips to the Philadelphia area for campaign events that are not far from his Delaware home. Last week, he traveled just west of the Philadelphia suburbs to Lancaster County, the heart of Pennsylvania Dutch country where Republicans have been losing their hegemony as the county’s population grows. There, Biden got in some licks of his own against the president, criticizing Trump’s response to the coronavirus by saying: “He’s like a child who just can’t believe this has happened to him. It’s all whining and self-pity.” In case anyone had not yet noticed, this year’s fight for the battleground state of Pennsylvania is on.

What follows are a series of tables exploring the recent voting in Pennsylvania (Tables 1-8) as well as Map 1, which shows how the state’s presidential voting patterns have changed over the last three elections.

Table 1: States decided by 2% or less in 2016

Table 2: Winning Pennsylvania since 1960

Table 3: PA cities, suburbs, and rest of state since 2000

Sources: America at the Polls 1960-2004: John F. Kennedy to George W. Bush (CQ Press). The 2008, 2012, and 2016 editions of America Votes (CQ Press, an imprint of SAGE) for presidential elections since then.

Table 4: PA counties that switched parties in 2012, 2016

Sources: America Votes 28, 30, 32 (CQ Press, an imprint of SAGE).

As Map 1 shows, down to the township level, the Obama areas that Trump flipped in 2016 were mostly found in the northeast and southwest, with their history of mining, as well as in Erie County, at the northwestern edge of the state. In the southeast, though, Chester County was the Romney > Clinton capital of the state.

Map 1: Pennsylvania partisan loyalty by town, 2008-2016

Table 5: PA GOP ballots up in ’16, Democratic vote down

Table 6: Voter registration trends since 2000

Source: Party registration totals are from the web site of the Pennsylvania secretary of state, with the latest as of June 15, 2020.

Table 7: Voter registration trends since Nov. 2016

Table 8: The Weld factor

Sources: Republican primary vote totals are from the websites of election authorities in states that held Republican presidential primaries in 2020 and William Weld received votes. The statewide victory margins from 2016 are from America Votes 32 (CQ Press, an imprint of SAGE).

KEY POINTS FROM THIS ARTICLE

— Unlike in 2016, Bernie Sanders has a real chance to win the Democratic presidential nomination.

— However, he likely will have to broaden his base of support to do so.

— Namely, better showings in big urban and suburban areas are important, particularly as the field narrows.

Sanders 2016: A look back

Bernie Sanders begins his second bid for the Democratic presidential nomination in possession of something he never attained in 2016: A competitive chance of winning.

Sanders’ first try four years ago was respectable. Facing a top-heavy favorite in Hillary Rodham Clinton, he won 22 states — 12 caucuses and 10 primaries, among them the battleground states of Michigan and Wisconsin. He drew 43% of the nationwide Democratic primary vote, which represented more than 13 million voters. As a result, he posted the highest primary vote total in the nation’s history for any candidate not named Obama, Clinton, or Trump.

Yet in 2016, Sanders never had a realistic chance of winning the party’s nomination. Two basic stumbling blocks stood in his way: superdelegates and the South. The former, which comprised 15% of the convention delegates, went virtually en masse for Clinton, as she was a part of the Democratic establishment in a way that Sanders never was or could be. And with Clinton’s firm grip on the minority vote, the Vermont senator was never able to penetrate the South. He lost 12 of 13 primaries across the region (all save Oklahoma), polling barely one third of its aggregate primary vote in the process.

Sanders’ problem garnering the votes of African Americans and Hispanics extended to other regions of the country as well, helping Clinton to dominate the vote in many of the nation’s leading urban centers and their suburbs. The result: In the 10 states with 15 or more electoral votes, Sanders could carry the primary in only one, and that, Michigan, was by less than 20,000 votes out of 1.2 million cast.

Basically, the heart of Sanders’ coalition in 2016 was academic centers and a significant swath of rural America. The latter was an unlikely source of votes for a self-described “democratic socialist.” Antipathy to Clinton was no doubt a major reason for his strong rural vote. So were his full-voiced attacks on what he saw as an insensitive political and economic elite. And in spite of his New York accent, his base in rural Vermont gave him a connection to primary voters in smaller states that Clinton could not match.

Of the 10 primaries that Sanders won, there were three types: those with a progressive pedigree (such as Oregon, Vermont, and Wisconsin); those that were New England neighbors of Vermont (New Hampshire and Rhode Island); and a mixed band of others (from Michigan and Montana to the unlikely trio of Indiana, Oklahoma, and West Virginia). Sanders also had close losses of two percentage points or less in the Iowa caucuses and primaries in Illinois, Kentucky, Massachusetts, Missouri, and South Dakota. He ran particularly well in states where independents were allowed to vote in the Democratic primary.

Sanders’ strength in rural areas was evident in the number of counties he carried in a variety of primary states outside the South. In Wisconsin, he won 71 of 72, losing only Milwaukee County to Clinton. In Oregon, he swept 35 of 36, losing only one small county to Clinton by a vote of 101 to 100. In Oklahoma, Sanders carried 75 of 77 counties. And in Michigan, he took all but 10 of the state’s 83 counties.

His victory in the Wolverine State was a microcosm of his strengths and weaknesses in 2016. Clinton dominated the Detroit metro area, winning Wayne County, which includes the city of Detroit, by 60,000 votes. She also carried the city’s two major suburban counties, Oakland and Macomb, the latter the fabled home of blue-collar “Reagan Democrats.” Outside the Detroit area, Clinton picked off Genesee County, which includes Flint, an industrial outpost that is the birthplace of filmmaker and progressive activist Michael Moore (a Sanders backer). But Sanders swept most everywhere else in Michigan, including the county of Washtenaw, which includes the University of Michigan at Ann Arbor and nearby Eastern Michigan University in Ypsilanti.

To be sure, the 2020 Democratic nominating race has a different complexion to it than that of four years ago. Then, Sanders was engaged in a one-on-one fight with Clinton where he needed a majority of the vote in primary and caucus states to prevail.

This time, that will not be the case, at least in the early voting. Pluralities will do, as the Democrats have a far wider field of candidates, including two billionaires in Michael Bloomberg and Tom Steyer, whose wealth gives them staying power. At this point, it would be no surprise if the wide field of Democratic candidates persisted well into the glut of March primaries. Sanders’ ardent group of supporters, augmented by his ability to raise impressive sums of money, should keep him as a major player in the Democratic race provided he does not unexpectedly tank in Iowa and New Hampshire.

Still, to win the Democratic nomination in 2020, Sanders will ultimately need to be more than the Democratic champion of academe and the rural countryside. Maybe not at the start, but eventually, he will need to show broader vote-getting ability than he did in 2016. That includes breakthroughs in the cities and the suburbs, both critical to Democratic success in the general election, and blue-collar jurisdictions with an industrial heritage such as Flint.

This year’s Democratic race could be profoundly affected by rules changes instituted by the party for 2020. Probably the most significant redefines the controversial role of superdelegates, which are party and elected officials given automatic delegate status by virtue of their position. Superdelegates are free to vote for the candidate of their choice regardless of the result of their state’s primary or caucus.

In 2016, they were crucial to the nomination of Clinton: She would have won without them, but their backing reinforced her edge. This time they will have no vote at all on the first ballot unless there is already a clear-cut winner going into the convention. It is a change that should work to Sanders’ advantage.

But another rules change may not. It encourages states to select their delegates through higher turnout primaries rather than comparatively low turnout caucuses. The latter are a venue that rewards passion, and Sanders’ enthusiastic supporters gave him victories in 12 of the 14 caucus states in 2016, often by lopsided margins. Iowa and Nevada will hold caucuses in 2020, but there will be few other states using the caucus process.

What follows is an in-depth examination of Sanders’ 2016 performance. Table 1 shows the difference in Sanders’ showing in caucuses versus primaries; Table 2 shows the primary vote by region; and Table 3 analyzes a series of sample counties, illustrating Sanders’ strengths and weaknesses in different kinds of places across the country.

Table 1: Sanders dominated in caucuses, Clinton in primaries

Note: The total number of states won by Sanders and Clinton totals 53, because the District of Columbia primary is included as are both caucuses and non-binding primaries in Nebraska and Washington. In the latter two, Sanders won the caucuses while Clinton took the non-binding primaries. A tally of actual caucus votes was taken in only eight of the 14 states that held them. In the other six caucus states, other measurements of candidate support were used.

Sources: America Votes 32 (CQ Press, an imprint of SAGE) for 2016 Democratic primary results; CNN The Republican and Democratic National Conventions Research & Editorial Guide 2016 (Robert Yoon, editor) for 2016 Democratic caucus results.

Table 2: Regional breakdown of the 2016 Democratic primary and delegate count

Note: The delegate count is based on the presidential roll call at the 2016 Democratic National Convention in Philadelphia. An asterisk (*) that accompanies Nebraska and Washington indicates that the states held a non-binding Democratic primary, with delegates selected in a separate caucus process. The District of Columbia is included in the list of Democratic primaries.

Source: America Votes 32 (CQ Press, an imprint of SAGE), 54-55, for the 2016 Democratic primary vote. The delegate tally is based on the 2016 Democratic convention presidential roll call and was compiled by the author, with assistance from the Frontloading HQ website and Robert Yoon of CNN.

Table 3: 2016 Democratic primary results: A sampling of counties

TYPES OF COMMUNITIES
At its most elemental level, the nation can be divided into three separate types of communities: urban, suburban, and rural. The largest cities are the cornerstone of the Democratic vote. Rural and small-town America are decidedly Republican. The suburbs are often up for grabs and swing between one party and the other, frequently deciding elections in the process. In the 2016 Democratic primaries, Clinton dominated major urban centers with their large minority populations, and tended to have the upper hand in suburban counties as well. As for Sanders, he swept many rural counties, especially outside the South.

Major urban centers

Major suburban counties

Rural and small-town America

RACIAL COMPOSITION
Minorities, especially African Americans, are arguably the most loyal element of the Democratic coalition. And counties with a sizable African-American population consistently gave Clinton her highest proportion of the Democratic primary and general election vote in 2016. The counties with the highest percentage of African Americans can generally be found in the South; those with the highest proportion of Hispanics in Florida and the Southwest. The minority percentages listed below are based on the 2010 census.

African-American majority

Hispanic majority

LIBERAL BASTIONS
One of the biggest surprises from the 2016 Democratic presidential primaries, at least in this corner, was that Bernie Sanders did not carry San Francisco. Yet he did dominate in other liberal strongholds, especially academic communities. An asterisk (*) indicates that the county or town is the home of multiple colleges. Amherst, Mass., is the site of three colleges: the University of Massachusetts, Amherst College, and Hampshire College. Washtenaw Co., Mich., is the home of the University of Michigan in Ann Arbor and Eastern Michigan University in neighboring Ypsilanti.

College towns

Urban centers

Artistic communities

MORE CONSERVATIVE VENUES
It was not so long ago that labor unions were strong and Democrats were champions of the working class. But in 2016, Donald Trump penetrated deeply into the blue-collar vote, which proved to be a major factor in his upset victory over Hillary Rodham Clinton. Democrats hope to be able to effectively compete with Trump for blue-collar support in 2020, but Sanders will have to make a better case than he did the last time that he is the candidate that can do it.

Industrial heritage

Military influence

Obama-to-Trump counties

Conclusion

New rules and a new field of candidates can make for a changing coalition for a second-time candidate such as Sanders. In the 2008 Democratic primaries, for instance, Clinton lost the African-American vote to Barack Obama. In 2016, she dominated it against Sanders. This time, it is Sanders’ turn to see if he can expand his coalition from last time. To put himself on the road to the White House in 2020, he will need to.

KEY POINT FROM THIS ARTICLE

— Altogether, there are 31 states (plus the District of Columbia) with party registration; in the others, such as Virginia, voters register without reference to party. In 19 states and the District, there are more registered Democrats than Republicans. In 12 states, there are more registered Republicans than Democrats. In aggregate, 40% of all voters in party registration states are Democrats, 29% are Republicans, and 28% are independents. Nationally, the Democratic advantage in the party registration states approaches 12 million.

Poring over party registration

This is not the best of times for the Democratic Party. No White House; no Senate; no House of Representatives; and a clear minority of governorships and state legislatures in their possession. Yet the Democrats approach this fall’s midterm elections with an advantage in one key aspect of the political process — their strength in states where voters register by party.

Altogether, there are 31 states (plus the District of Columbia) with party registration; in the others, such as Virginia, voters register without reference to party. Among the party registration states are some of the nation’s most populous: California, New York, Florida, North Carolina, Pennsylvania, New Jersey, Arizona, and Massachusetts.

The basic facts: In 19 states and the District, there are more registered Democrats than Republicans. In 12 states, there are more registered Republicans than Democrats. In aggregate, 40% of all voters in party registration states are Democrats, 29% are Republicans, and 28% are independents. Nationally, the Democratic advantage in the party registration states approaches 12 million.

Still, Republican Donald Trump found a route to victory in 2016 that went through the party registration states. He scored a near sweep of those where there were more Republicans than Democrats, winning 11 of the 12, while also taking six of the 19 states where there were more Democrats than Republicans — a group that included the pivotal battleground states of Florida, North Carolina, and Pennsylvania.

Map 1 and Table 1: Party registration totals by state, July 2018

Sources: Recent party registration numbers used here are from state election websites and are based on totals compiled in early July 2018. Registration data are as of the following months: October 2016 (Kansas and Rhode Island); February 2017 (Massachusetts); November 2017 (Connecticut); January 2018 (Oklahoma); March 2018 (Arizona and Maine); April 2018 (New Hampshire and New York); May 2018 (California, Florida, Maryland, Nebraska, and Pennsylvania); June 2018 (Arkansas, District of Columbia, Kentucky, New Jersey, New Mexico, North Carolina, and West Virginia); and July 2018 (Alaska, Colorado, Delaware, Idaho, Iowa, Louisiana, Nevada, Oregon, South Dakota, Utah, and Wyoming).

Notes: In states such as Arizona, Colorado, Connecticut, Iowa, Maine, Nevada, New York, Rhode Island, South Dakota, and Utah, the total number of active and inactive voters are presented on the state election website, but only the number of active registered voters is used in this article. The “other” column includes voters registered with third parties. Percentages do not always add to 100 due to rounding. Zero percent (0%) indicates less than 0.5%. “NA” indicates there were no numbers in this category.

Some caveats

At this point, it might be wise to pause and ask the question: Why do these numbers matter, either individually or in the aggregate?

Certainly there are facts that argue that they should be taken with a grain or two of salt. Most party registration states are found in more Democratic terrain: the Northeast (11 states plus the District of Columbia) and the West (10 states), followed by the South (six states) and the Midwest (four states), all of the latter rural states west of the Mississippi River.

To be sure, there are a number of major states that do not register voters by party, such as Texas, Georgia, Washington, and the keystones of the industrial Midwest: Illinois, Indiana, Michigan, Ohio, and Wisconsin. If they did register by party, Texas, Georgia, and Indiana would almost certainly add to the Republican total; the industrial states probably less so.

Comparing party registration totals between states at a particular point in time can be a bit misleading, particularly because some may have recently completed a purge of “dead wood” on their rolls while others have not. Often, the trend line over a series of years is more valuable to look at than a one-time registration count.

States can also differ on how they report active versus inactive voters. Basically, the latter have not voted in several consecutive elections and have not responded to the efforts of election officials to reach them, though they are generally allowed to show up and vote if they have not moved. Some states fold in active with inactive voters to produce one grand total. A few states break the two categories out separately, where the number of active voters routinely dwarf the number of inactive voters, sometimes by a ratio in excess of 10 to 1. (This article uses only state totals of active voters, wherever possible.)

And there is some sentiment that a voter’s party identification may mean less than it once did, as the number of individuals who register as “Independent” (or “No Party Preference,” “Unaffiliated,” or whatever other nom de plume the individual states prefer) steadily grows. At the beginning of this century, barely 20% of all voters in party registration states were independents. Nowadays, that total is approaching 30%.

Altogether, there are 10 states with more registered independents than either Democrats or Republicans. These states are mainly in the Northeast, with a cluster also in the West. By comparison, there are Democratic pluralities of registered voters in 13 states plus the District of Columbia and eight other states with Republicans ahead of both Democrats and independents. In addition, there are six states where there is an independent plurality but Democrats outnumber Republicans, and four states where independents are on top of the registration totals but Republicans outnumber Democrats. That produces the 19 to 12 state registration advantage for the Democrats mentioned earlier.

With the growth in independents, many voters seem to be saying to the two major parties: “a pox on both your houses.”

Yet it also can be argued that registering Democratic or Republican is far more of a statement than it once was. In the current age of sharp-edged partisanship, there is far more than a “dime’s worth of difference” between the two major parties, so registering as a Democrat or Republican is a very intentional act of differentiation.

And that makes the party registration figures worth looking at. A comparison of party registration totals on the eve of the 2016 presidential election with the actual voting in November shows a noticeable correlation between party registration and the state by state election outcomes. Twenty-four of the 31 party registration states were won by the nominee whose party had more registered voters (discounting independents for this particular comparison). That is a 77% correlation rate between party registration advantage and a winning electoral outcome. The percentage goes up to 88% if one removes the South, the one area of the country where party registration is a lagging indicator of the fortunes of the two major parties.

Chart 1 and Table 2: Nationwide party registration trends since 2000

Note: Based on active registered voters in states where the number of active and inactive registrants is listed. In the election-eve 2000, 2008, and 2016 entries, “Independents” include a comparatively small number of registered miscellaneous voters who do not fit into a particular category. Percentages do not add to 100 since the small percentage of registered third party voters is not included.

Sources: Richard Winger’s monthly newsletter, Ballot Access News, for election-eve party registration numbers in 2000, 2008, and 2016; the websites of state election offices for July 2018.

Southern exceptionalism

Even though Democrats began losing their dominance in the South a half century ago, they still retain a registration advantage in four of six Southern states where voters register by party: Florida, Kentucky, Louisiana, and North Carolina.

In Arkansas, there is a rudimentary system of party registration in which there are a few more Republicans than Democrats, but a vast majority of voters do not identify with either party. In Oklahoma, one of the reddest of states over the last few decades, the party registration advantage did not finally flip from Democratic to Republican until a few years ago during Barack Obama’s presidency. (Oklahoma is the only state this century where the registration advantage has switched between parties and the Republican lead is likely to hold for the long term. In battleground states such as Colorado, Iowa, and Nevada, the registration advantage has switched at least once between parties since 2000 and has less of a look of permanence.)

It is an open question as to why other party registration states in the South have not also flipped to the Republicans. For years, they have been regularly voting for GOP candidates up and down the ballot, including 2016 when the whole region except Virginia went for Trump. White southerners who once registered en masse for the Democrats have been dying off for years. And the time when voters registered Democratic because its primary was tantamount to election is long gone. Still, in booming states such as Florida and North Carolina, an influx of outsiders registering Democratic could be helping the party fend off a large GOP registration surge that would flip the states to the Republicans.

As it is, Republicans have been steadily whittling away at the Democrats’ registration advantage in the Southern party registration states. In Florida, the GOP deficit is now less than 230,000 registered voters after being nearly 660,000 in October 2008. In North Carolina, Republicans trail by less than 575,000 registered voters, after being down by 865,000 a decade ago; while in Kentucky and Louisiana, the GOP registration deficit nowadays is below 300,000 and 400,000 registrants, respectively, less than half of what it was in each state on the eve of the 2000 election.

Table 3: State-by-state registration trends since 2000

Notes: The largest Democratic and Republican registration margins among the four data points are in bold. An asterisk (*) indicates that in July 2018, there were more voters registered as independent than Democratic or Republican. “NA” indicates that party registration figures are not readily available. Party registration totals are based on the number of active voters in states where totals for both active and inactive voters are posted.

Sources: Election-eve party registration figures for 2000, 2008, and 2016 were compiled by Richard Winger and published in his monthly newsletter, Ballot Access News. The July 2018 data were compiled by the author from party registration numbers posted on state election websites (see Sources note for Map 1 and Table 1).

Lots of consistency elsewhere

In the rest of the country, there was much more consistency between party registration totals and the 2016 election outcome, with only three non-Southern states voting “against the grain.” On election eve in Pennsylvania, there were 915,081 more registered Democrats than Republicans; Trump carried the state by 44,292 votes. In West Virginia, there were 175,867 more registered Democrats; Trump won by 300,577 votes. And in New Hampshire, there were 24,232 more registered Republicans than Democrats in the fall of 2016, but Hillary Clinton took the state by 2,736 votes. That’s it. The other 22 party registration states outside the South were carried in the presidential balloting by the party with more registered voters than the other.

And in many of these “in sync” states, the registration advantage in recent years has grown more Republican or Democratic as the case may be, augmented by a healthy increase in independents.

Case in point: California. Eighteen years ago, 45% of the state’s 15.7 million registered voters were Democrats, 35% were Republicans, and 14% were independents. Last month, on the eve of California’s June 5 primary, 44% of the 19.0 million registered voters were Democrats (down one percentage point from 2000), 25.5% were independents (up 11 points), and 25% were Republicans (down 10 points), as the Republicans dropped to third place in California registration totals and independents rose to second. In the process, the nation’s most populous state has gone from one that was vaguely competitive for the Republicans at the beginning of the 21st century to one where Republicans have trouble competing statewide at all.

The registration trend line in California is a microcosm of sorts of party registration in the nation as whole. Democrats are running ahead and the ranks of the independents are growing. Yet registered voters in both parties appear to be widely engaged. That was the case in 2016, and likely will be again in 2018, with Trump flogging issues to rouse his base. In short, this is a highly partisan era when party registration totals, and the trends that go with them, are well worth watching.

Map 2 and Table 4: Party registration and the 2016 presidential vote

Notes: An asterisk (*) indicates states where there were more registered independents than either Democrats or Republicans in October 2016. “Independents” include a comparatively small number of registered miscellaneous voters who do not fit into any particular category.

Sources: Richard Winger’s Ballot Access News for October 2016 party registration data; America Votes 32 (CQ Press, an imprint of SAGE) for 2016 presidential election results.

KEY POINTS FROM THIS ARTICLE

— Although Donald Trump is remaking the Republican Party in his image, he had among the shortest coattails of any presidential winner going back to Dwight Eisenhower. In 2016, Trump ran ahead of just 24 of 241 Republican House winners and only five of 22 Republican Senate winners.

— While more Republican House members are from the South than any other region, Trump’s coattails were longest in the Midwest, where he ran ahead of nine Republican House winners. Trump ran ahead of eight victorious GOP House candidates in the South, and a combined total of seven in the two Democratic bailiwicks, the Northeast and the West.

— The length of presidential coattails over the years has been closely related to the winner’s share of the popular vote. Landslide victories (say, with 55% or more) tend to produce much longer coattails than those of “minority” winners such as Trump. But in the eight presidential elections from 1988 to 2016, there has been nary a landslide and the length of presidential coattails has shrunk.

Measuring Donald Trump’s coattails

It is one of the ironies of modern American politics that congressional Republicans have bound themselves virtually en masse to a president with some of the shortest coattails in a generation.

For better or worse, Donald Trump has immersed himself in GOP politics — from fundraising to endorsing (and opposing) candidates to holding his patented campaign-style rallies for his favorites. But offering coattails for Republican candidates? Well, not so much.

The only time Trump has had a chance to exhibit his coattail pull came on Nov. 8, 2016, when he was elected with a solid 304 electoral votes but only 46.1% of the popular vote — 2.1 percentage points behind Democrat Hillary Clinton. Of the 241 Republican House candidates elected at the same time, Trump outran just 24 of them — or put another way, one out of 10. It was the smallest demonstration of coattail pull by a presidential winner since 1992, when Bill Clinton’s 43% of the vote in a three-way race with President George H.W. Bush and Ross Perot enabled him to run ahead of just five House Democrats.

Some might argue that Trump’s brash, anti-establishment campaign struck a chord with many voters and helped set the tone for Republican victories in 2016. But when it comes to the quantitative side of presidential coattails, the vast majority of Republican House members owe Trump virtually nothing.

The same goes for the Senate, where Trump ran ahead of just five (of 22) Republican Senate winners in 2016. All of the Trump-led senators were in red America: Roy Blunt of Missouri, John Boozman of Arkansas, Lisa Murkowski of Alaska (who was involved in a three-way race), Rand Paul of Kentucky, and Todd Young of Indiana.

However, in swing states where control of the Senate was decided, Trump consistently drew a lower percentage of the vote than the Republican winner. He trailed Pat Toomey in Pennsylvania by six-tenths of a percentage point, Richard Burr in North Carolina by 1.3 points, Marco Rubio (his erstwhile GOP sparring partner) in Florida by three points, Ron Johnson in Wisconsin also by three points, John McCain in Arizona by five points, and Rob Portman in Ohio by fully six percentage points.

Table 1 and Map 1: Trump’s Modest Senate Coattails in 2016

Note: Each senator’s current term is based on full terms in office. An asterisk (*) indicates that the senator served an unfilled portion of a previous term before winning a full term. Percentages are based on each candidate’s share of the total vote. A Republican Senate victory by John Kennedy in Louisiana is not included in the table because it was not secured until a December runoff when Donald Trump was not on the ballot.

Source: America Votes 32 (CQ Press, an imprint of SAGE).

Yet when discussing presidential coattails, the focus is usually on the House — one, because Congress’ lower chamber has historically been more susceptible to the ebb and flow of national elections; and two, because in presidential election years, the House is the only other office contested in all 50 states.

To be sure, the quantitative measurement of presidential coattails has its limitations. The count includes all congressional districts in which the presidential winner drew a higher percentage of the vote than his party’s House winner, regardless of whether help from the top of the ticket could be considered crucial to victory or not. The list of Trump’s coattails, for example, includes Jason Smith of Missouri, who was reelected in 2016 with 74% of the vote at the same time that Trump was carrying his district with 75%. That counts as a presidential coattail, even though no one would seriously make the argument that Smith needed Trump’s help to win election.

A more meaningful tally of coattails would probably be limited to districts in which the presidential winner ran ahead of his party’s marginal House winners, defined here as those who drew less than 55% of the total vote. In 2016, there were six Republican House marginal winners who ran behind Trump: Martha Roby (AL-2), Don Young (AK-AL), Mike Bost (IL-12), Trey Hollingsworth (IN-9), Jack Bergman (MI-1), and Claudia Tenney (NY-22). Still, the count of presidential coattails traditionally uses the broader definition of the number of districts in which the presidential winner outruns his party’s House winners, regardless of the percentage.

In 2016, the “Trump 24” was largely made up of “backbenchers” unknown to the average voter. A dozen of those that he outpaced were relative newcomers to Congress, currently in their first or second terms — a number that included Liz Cheney, the daughter of the former vice president. She was elected to Wyoming’s open at-large House seat in 2016 with 62% of the vote as Trump was sweeping the state with 68%.

Table 2a and 2b: The “Trump 24,” Republican House winners who ran behind Trump in 2016, and Trump’s performance vs. GOP House leaders

Notes for Tables 2a and 2b: “Current term” is based on the number of full terms. Districts are not included as coattails where the vote share received by Trump and the Republican House winner was the same. Cases in point, IN-2, where Trump and Rep. Jackie Walorski each drew 59.3% of the vote, and MT-AL, where Trump and Rep. Ryan Zinke each received 56.2% of the vote. Trump coattails also do not include the two Louisiana districts where the Republican House victory was not settled until the December 2016 runoff.

Sources for Tables 2a and 2b: The 2016 presidential vote by congressional district was compiled by Daily Kos Elections, and was used by the author to derive the extent of Donald Trump’s coattails in 2016. The 2016 House vote is from America Votes 32 (CQ Press, an imprint of SAGE). The number of terms served by House members is from The Almanac of American Politics 2018 (Columbia Books & Information Services, National Journal).

In contrast, Republican House leaders routinely outpaced Trump. House Speaker Paul Ryan (WI-1) was reelected in 2016 with 65% of the vote, 12 percentage points higher than Trump’s showing in his district. House Majority Leader Kevin McCarthy (CA-23) won with 69%, 11 points ahead of Trump. And House Majority Whip Steve Scalise (LA-1) took 75%, running six percentage points better in his district than the GOP standard-bearer. Even Devin Nunes (CA-22), who has been criticized as a lackey of the president as chairman of the House Intelligence Committee, far outperformed Trump in his California district. The tally: Nunes, 68%; Trump 52%.

The examples of presidential coattails in 2016 often tended to involve congressional districts that were geographically clustered and rural in character. Three Trump-led districts were in Michigan. Another trio were in the Appalachian coal country of West Virginia and the adjacent Virginia panhandle (the “Fighting Ninth”). A pair of Trump coattails was in southern Indiana; another two were sprawled across southern Missouri; and two more were in Middle Tennessee.

Far more common, though, was the absence of Trump coattails in large Republican delegations. He outran none of the 10 GOP House winners in Georgia, none of the 10 in North Carolina, none of the 12 in Ohio, none of the 14 in California, and none of the 25 in Texas.

Tables 3a and 3b: Trump and the GOP House in 2016: A national comparison and Trump’s coattails by region

Notes for Tables 3a and 3b: Percentages in the far right column of the regional chart that ended in .500 are rounded up to the nearest whole percentage point. This analysis is based on the old Pennsylvania House map, not the new one that is likely to be in place for the 2018 election. The regions are defined as follows:

South (13 states) — Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Virginia.

Midwest (12 states) — Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, Wisconsin.

West (13 states) — Alaska, Arizona, California, Colorado, Hawaii, Idaho, Montana, Nevada, New Mexico, Oregon, Utah, Washington, Wyoming.

Northeast (12 states) — Connecticut, Delaware, Maine, Maryland, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont, West Virginia.

Sources for Tables 3a and 3b: The 2016 nationwide presidential and House votes are from America Votes 32 (CQ Press, an imprint of SAGE). The presidential vote by congressional district was compiled by Daily Kos Elections. The author used the latter data to compute the extent of Donald Trump’s congressional coattails in 2016 by region.

The length of presidential coattails over the years has been closely related to the winner’s share of the popular vote. Landslide victories (say, with 55% or more) tend to produce much longer coattails than those of “minority” winners such as Trump.

In most of the 20th century, landslides were the norm. Nearly half of the presidential elections from 1900 through 1984 — 10 of 22 — were won with at least 55% of the vote. And with the landslide victories came some very lengthy coattails. In 1956, Republican Dwight D. Eisenhower won reelection with 57% of the vote and in the process ran ahead of more than 150 of his party’s successful House candidates. In 1964, Democrat Lyndon B. Johnson rolled to a 61% victory and outpaced more than 130 of his party’s congressional winners. In 1972, Republican Richard Nixon scored a 61% landslide win and outran more than 100 of his party’s House victors.

But in the eight presidential elections since Ronald Reagan’s 49-state triumph in 1984, there has been nary a landslide and the length of presidential coattails has shrunk. Over the last three decades, most of the presidential winners have had coattails numbering in the 20 to 40 range. One has been below 20 (Clinton’s five in 1992); two have been above 40, paced by Barack Obama, who drew a higher share of the vote than 61 successful Democratic congressional candidates in 2012.

While long coattails may no longer be in their quiver, presidents still have ways to influence national campaigns. They can use the “bully pulpit” to fashion campaign themes, something that Trump is trying to do now for 2018. He is also raising money, endorsing Republican candidates (most recently Senate aspirant Mitt Romney in Utah), and seeking to cajole and intimidate both friends and foes alike.

With the Gallup Poll showing the president’s mid-February approval rating to be above 85% among Republicans, Trump will likely have his greatest impact this year in molding the party’s message and helping to define the list of nominees that Republicans will put before the voters in November. But with a Gallup approval rating of just 30% among independents, and barely 5% among Democrats, his role in the 2018 general election looks to be problematic.

There is little doubt that the controversial Trump will be the central player of the 2018 campaign. Even while his name is not on the ballot, this year’s elections will offer a highly charged referendum on Trump and his presidency. Love him or hate him, it could result in a midterm variation of presidential coattails, or a lack thereof.

Chart 1 and Table 4: Presidential coattails since 1956

Note: An asterisk (*) designates an incumbent president. Percentages in the far right column that ended in .500 are rounded up to the nearest whole percentage point.

Sources: The winning share of the total popular vote for president in each election since 1956 is from America Votes 32 (CQ Press, an imprint of SAGE). The number of House seats won by the president’s party is from the House clerk’s office. The number of congressional districts that the winning presidential candidate carried and the number in which he drew a larger share of the vote than a victorious House member of his own party is from Vital Statistics on Congress (updated September 2017) and covers elections from 1956 through 2012 (with the exception of 1968, a three-party election that Richard Nixon won with barely 43% of the vote). The data is found in Table 2-17 and has been posted on the Brookings Institution website. The 2016 presidential vote by congressional district was compiled by Daily Kos Elections and posted on its website. The author used the latter data to compute the extent of Trump’s congressional coattails in 2016.

If Hillary Clinton had won the presidency — and she took the popular vote by nearly 3 million — the narrative of the 2016 election would be far different. Rather than the storyline being Donald’s Trump triumph in the heartland, with its beleaguered blue-collar workers, the emphasis now would be on the Democrats’ ongoing success in metro America, with its large share of the nation’s growing minority population. The conventional wisdom would surely be that the Democrats were likely to control the White House for years to come.

That still might happen, and the 2016 presidential election may ultimately be viewed as an aberration that will be difficult for the Republicans to replicate. As it was, the election was a split decision — Clinton taking the popular vote, Trump the all-important electoral vote.

In the process, both candidates played to their party’s strengths, with an important caveat or two that made the 2016 election both different and noteworthy.

Clinton rolled up the vote in traditional Democratic bastions: major urban centers, high-tech areas, academic communities, minority strongholds (African American and Hispanic, in particular), and state capitals with their sizable government workforces. But she also extended the Democrats’ recent success in the suburbs, long a competitive sector of the electorate.

Trump, for his part, scored big in the usual Republican preserves: rural and small-town America, retirement communities, heavily white counties, and those with military installations. He also made important inroads in historically Democratic industrial areas. Among the counties that he drew into the Republican column were Macomb County (MI), Luzerne County (PA), and Kenosha County (WI).

Macomb County, where reporters often go in search of blue-collar “Reagan Democrats,” has become a national bellwether of sorts, voting for Obama both times he ran for president and Republican George W. Bush in 2004. In contrast, Luzerne County had not bolted from the Democrats since 1988, nor Kenosha County since Richard Nixon trampled George McGovern in 1972. All three counties, plus a number of other blue-collar ones that Trump carried, were critical to his cracking the Democrats’ supposedly impregnable “blue wall” in Michigan, Pennsylvania, and Wisconsin.

As for Clinton, she pushed the Democrats’ reach deeper into suburbia, flipping such counties as Cobb and Gwinnett outside Atlanta, Chester outside Philadelphia, and Orange in southern California from the Republican side in 2012 to the Democratic column in 2016.

Pennsylvania’s Chester County had voted for Barack Obama in 2008. But Cobb and Gwinnett counties had not voted for a Democrat since 1976, when Georgian Jimmy Carter first ran for president. And Orange County, the quintessential home of Sun Belt conservatism, last voted Democratic for president in 1936, at the height of Franklin D. Roosevelt’s New Deal.

Antipathy to Trump was no doubt a factor in the Democrats’ success in the suburbs, but so was a growing minority population in many of these counties. Orange County, for instance, which was barely 20% minority in 1980, was majority-minority by 2010.

Yet there was also a feast or famine quality to the Clinton and Trump votes. Many counties in rural America did not give Clinton even 25% of the vote, and sometimes far less than that. It helped to make her beatable in states that she was counting on winning.

On the other hand, in a number of the country’s major urban centers, Trump quite simply tanked. He captured barely one-third of the vote in Florida’s heavily Hispanic Miami-Dade County, and less than one-third of the vote in Ohio’s Cuyahoga County (Cleveland) and Michigan’s Wayne County (Detroit). In Los Angeles County, the nation’s most populous, Trump drew barely 20%, as was also the case in Cook County (Chicago) and Washington’s King County (Seattle). He took less than 20% in his home base of New York City, only 15% in Philadelphia, less than 15% in Boston, under 10% in San Francisco, and less than 5% in his new home, Washington, DC.

Yet in spite of Clinton’s dominance of the major urban centers, her margins were not quite big enough in the kingpins of Michigan (Wayne County), Pennsylvania (Philadelphia), and Wisconsin (Milwaukee County), to carry any of those three pivotal states. Clearly, she was hurt by a lack of enthusiasm for her candidacy among African-American voters, who had been much more eager to vote for one of their own in Obama.

Clinton won Wayne County by 290,000 votes, more than 90,000 votes less than Obama’s 2012 margin over Mitt Romney. She carried Philadelphia by a hefty 475,000 votes, but still trailed Obama’s margin four years earlier by 17,000 votes. And Clinton took Milwaukee County by nearly 163,000 votes, but again fell almost 15,000 votes short of Obama’s 2012 margin.

If she had come anywhere close to matching Obama’s 2012 margin in Wayne County, she would have wiped out Trump’s 10,704-vote statewide advantage and carried Michigan and its 16 electoral votes. However, even if Clinton had been able to match Obama’s 2012 margins in Philadelphia and Milwaukee County, it would not have been enough for her to overcome a 22,748-vote deficit in Wisconsin and a 44,292-vote deficit in Pennsylvania. In both states, she would have needed to make up more votes elsewhere.

A closer look at the Pennsylvania results illustrates the problem that Clinton faced. While she slightly underperformed in Philadelphia compared to Obama’s 2012 showing, she over-performed in the Philadelphia suburbs. She swept all four suburban counties (Bucks, Chester, Delaware, and Montgomery) by a combined margin of nearly 190,000 votes, 65,000 votes better than Obama’s suburban advantage four years earlier.

That gave Clinton a lead of more than 660,000 votes heading out of the Philadelphia metro area. Add an edge of nearly 110,000 votes for Clinton in Pennsylvania’s other major population center, Allegheny County (Pittsburgh), and she had an impressive lead of more than 770,000 votes — 65,000 votes larger than Obama’s in 2012 — to withstand a Republican counterattack in the rest of the state.

Obama had been able to do so easily, as Romney could trim only 400,000 votes from Obama’s lead in the vast swath of Pennsylvania outside the Philadelphia and Pittsburgh areas. But the Trump forces concentrated on the state’s interior, with its small and medium-sized cities, old industrial centers, small towns, and rural terrain. Much of the resource-rich territory had seen better days economically and was tailor-made for Trump’s pledge to “Make America Great Again.” While Romney carried seven counties in Pennsylvania with at least 70% of the vote, the number for Trump was 23.

The result was a Trump margin of more than 800,000 votes outside the Philadelphia and Pittsburgh areas, enough for a narrow statewide win. It was the first time that Republicans had carried the Keystone State’s electoral votes since 1988, and they did it with a different playbook than then. In the 1980s and before, the Republicans’ path to victory in Pennsylvania often started by building a big lead in the Philadelphia suburbs. Not so with Trump, who won both the state and the White House by following his own version of the path less traveled.

What follows is a look at how Clinton and Trump did, respectively, in different types of counties across the country. While 2016 was not that much different nationally than 2012 — the Democratic candidate won by about two points nationally instead of by four — there were wild swings at the state and county level that, taken together, allowed Trump to capture the presidency.

2016 results as seen through different types of communities

Group 1: Urban, suburban, rural

Table 1: Major urban centers

Note: *Obama won Harris County 49.4% to 49.3% in 2012.

Table 2: Major suburban counties

Table 3: Small-Town America

Group 2: The old and new economies

Table 4: Industrial heritage

Note: *Trump won Pueblo County 46.1% to 45.6% in 2016.

Table 5: High tech

Group 3: The young and the old

Table 6: Academic influence

Table 7: Retirement areas

Group 4: A look by race and ethnicity

Table 8: African-American majority

Table 9: Hispanic majority

Table 10: Almost monolithically white

Group 5: The left and the right

Table 11: State capitals

Table 12: Military influence