For the first half of the year, I was writing the state and gubernatorial chapters of the Almanac of American Politics 2018, the once-every-two-years, 2,000-page-plus "Bible of politics" that is being released in early August.
As part of that effort, I gave special attention to state- and county-level data on last year's presidential voting patterns, focusing on locations where Donald Trump and Hillary Clinton did well or poorly compared to their party's performance in 2012.
I found some interesting results. Let's start by looking at Trump.
Trump's Areas of Strength
In a number of states, Trump's candidacy resonated so strongly with voters that these places became even redder across the board than they were in 2012. These states included Indiana, Kentucky, Missouri and West Virginia.
In others, Trump didn't do well enough to win, but he did do well enough to cut into the Democrats' winning margins.
In Rhode Island, for instance, the winning Democratic margin shrank by 13 percentage points between 2012 and 2016. Similarly, Delaware saw its margin shrink by eight points; Minnesota, by about seven points; New Hampshire, by six points; and Nevada, about four points.
Two demographic groups, in particular, produced notable results for Trump's candidacy. The first is senior citizens, whose influence was most noticeable in Florida.
Trump ran strong in the Interstate 4 corridor that stretches from Tampa past Orlando. He flipped from blue to red such counties as St. Lucie and Pinellas, which includes St. Petersburg. He also saw six-digit raw-vote increases in the counties of Lee, home to Fort Myers; Pasco; Volusia (Daytona Beach); Polk; Manatee (Bradenton); Hernando; Sarasota; and Charlotte.
The strong affinity between Trump and non-college educated voters has been extensively explored since the campaign. But a state-by-state look offers stunning evidence of just how large the shifts were in certain counties.
One of the best examples of this is Colorado's Pueblo County, a historically Democratic stronghold that is home to many blue-collar, religious, gun-friendly voters. Barack Obama won the county by 14 points. This time, though, Trump won it, albeit narrowly, and even as incumbent Democratic Sen. Michael Bennet won re-election in the county by about 7,000 votes.
There were similar shifts in solidly blue states.
In Connecticut, the town of Plainfield, near the border with Rhode Island, went for Obama by 10 points but backed Trump by 22. Naugatuck, south of Waterbury, backed Trump by 16. In New York, Trump won 19 counties that had voted for Obama in 2012. Franklin County and Oswego County both saw their margins shift by more than 30 points toward the GOP. Five other counties saw margins shift by between 20 and 29 points, while another nine shifted by between 10 and 19 points.
But the more important shifts for Trump -- for his victory in the Electoral College, anyway -- came in competitive states.
In Maine, Trump won nine counties in the state, which was eight more than Romney did in 2012. It was enough to secure an electoral vote (Maine allocates a portion of its electoral votes by congressional district). The margin in Oxford County shifted 28 points toward Trump; it shifted 25 points in Aroostook County, 24 in Franklin County, 17 in Kennebec County (Augusta), and 14 in Penobscot County. And it wasn't just the margins that gave him a victory: In these five counties alone, the overall level of voter turnout rose by 6 percent over 2012.
In Ohio -- which went from blue to red in 2016 -- Trump's gains were especially strong in the state's blue-collar northern tier. In seven of the Obama counties that Trump was able to flip, the margins shifted toward the GOP by between 12 and 32 points. Even in some of the counties that remained blue, the Democratic margins narrowed significantly, shrinking by 25 points in Mahoning County (Youngstown), by 16 points in Lorain County (west of Cleveland), by 14 points in Lucas County (Toledo) and by seven points in Summit County (Akron). In many places, Trump outperformed Romney robustly, such as Stark County (Canton), where a virtual tie in 2012 became a 17-point Trump romp in 2016.
In Pennsylvania, the cumulative Republican improvement over 2012 in just five blue-collar counties in western Pennsylvania -- Beaver, Fayette, Greene, Washington and Westmoreland -- was almost enough by itself to supply Trump's statewide winning margin in this previously blue state. Trump's margin was also in the ballpark of his improvement in Luzerne County (Wilkes-Barre) and Lackawanna (Scranton).
In Wisconsin, another blue-to-red state, Clinton won only about one-third of the counties Obama had won four years earlier. The swing in Trump's direction reached 31 points in Forest County, which is located between Green Bay and Lake Superior. A Milwaukee Journal-Sentinel analysis found that in Wisconsin communities of less than 2,000 people, Trump won by 24 points, a margin six times bigger than Romney's. And while Trump lost the state's metro areas by five points, the newspaper calculated, he won non-metro areas by 19 points.
Visually, the most striking change on the map might have occurred in Iowa.
There, every county touching the Mississippi River voted for Obama in 2012, yet all but one switched to Trump last year. The swing in these counties was frequently huge: Jackson County shifted by 37 points; Lee County, by 32; Clinton County, by 2; and Muscatine County, by 22. Changes of similar magnitude also occurred in counties on the Illinois side of the Mississippi River. For instance, Henderson County, Ill., shifted by 40 points toward Trump.
What About Clinton?
Now, let's turn our attention to how things shook out for Clinton. In several states, she triumphed essentially across the board.
In California, for instance, her raw vote total was 11 percent higher than Obama's in 2012, while Trump's total fell by 7 percent from Romney's. In each of California's seven most populous counties, Clinton improved on Obama's 2012 raw vote totals, by between 36,000 (San Bernardino) and 247,000 votes (Los Angeles). And in Massachusetts, Clinton won every county by at least 7 percentage points, and she exceeded Obama's vote tally by almost 73,000, while Trump fell 97,000 votes short of Romney's total.
Heavily Latino Areas
More commonly, however, Clinton's gains -- not always wins, to be sure, but at least improvements over Obama in 2012 -- can be credited to certain demographic characteristics.
In Arizona's populous Maricopa County, Clinton came within about 40,000 votes of defeating Trump, down from Romney's winning margin of 148,000 in 2012. Latinos there turned out in droves to vote against Sheriff Joe Arpaio, who was also on the ballot. In Florida, Clinton won Hispanic votes by 62 percent to 35 percent, and secured six-digit raw-vote increases in the Democratic bastions of Miami-Dade, Broward (Fort Lauderdale), Palm Beach and Orange counties, all of which have sizable Hispanic populations.
In Virginia, such ethnically diverse Democratic jurisdictions as Arlington, Fairfax, Loudoun and Prince William counties, as well as the city of Alexandria, saw their winning margins shift by between five and 20 points in the Democrats' direction.
And in Texas -- now a majority-minority state -- Clinton's tally increased by about 570,000 votes between 2012 and 2016, far more than Trump's increase of 115,000 in the state. Notably, Fort Bend County near Houston -- the state's most racially and ethnically diverse -- shifted sharply toward Clinton. The county backed Romney by a seven-point margin in 2012, but it flipped in 2016 to support Clinton by six points.
By contrast, some heavily African-American areas saw support for Clinton decline compared to Obama.
For instance, in Maryland, Clinton saw a five-digit raw-vote decline in heavily Democratic and minority Baltimore city. In Ohio, she tallied 49,000 fewer votes in Cuyahoga County (Cleveland). And in Michigan's Wayne County (Detroit), Clinton's raw-vote total dropped by 76,000 and her winning margin shrank by 10 points. In Genesee County, which includes Flint, it shrank by 20 points and Democratic raw votes dropped by one-fifth.
Meanwhile, in Mississippi -- the state with the highest percentage of African-Americans in the country -- Clinton saw her statewide votes decline by 78,000 compared to Obama's total from four years earlier.
Democratic strength in the suburbs was evident in blue states and red states alike.
In California, Orange County -- which hadn't voted Democratic for president since Franklin Roosevelt in 1936 -- swung 15 points in the Democratic direction in 2016. In Connecticut, affluent Darien had backed Romney by 31 points, but Clinton won it by 12 -- a 43-point swing. In neighboring New Canaan, the swing toward Clinton was 42 points, and in Greenwich, it was 29 points.
In the Chicago suburbs of Illinois, four of the five collar counties (DuPage, Lake, Will and Kane) saw the margin shift in the Democratic direction by as much as 13 points between 2012 and 2016. And in the three core Oregon counties around Portland -- Multnomah, Washington and Clackamas -- Clinton improved upon Obama's 2012 showing by 42,000 votes.
Interestingly, the same pattern also emerged in red states.
Metro Atlanta serves as a prime example. Three counties near Atlanta shifted from Romney to Clinton: Gwinnett, with a 17-point swing toward Clinton; Cobb, with a 14-point swing; and Henry, with a seven-point swing. In addition, Clinton carried several metro counties, including Fulton, Douglas, Rockdale and DeKalb by margins that were six to 13 points higher than Obama.
Such shifts were detectable even in states redder than Georgia.
In Kansas' Johnson County, a populous suburb of Kansas City, Romney won by 17 percentage points, but Trump won by fewer than 3; Trump underperformed Romney in the county by 21,000 votes while Clinton overperformed Obama by 19,000. And in Oklahoma, the GOP margin in Tulsa County dropped from 28 points to 22; in Oklahoma County (Oklahoma City), it fell from 16 to 11; in Canadian County (Oklahoma City suburbs), from 54 to 51; and in Cleveland County (Norman), from 26 to 21.
There were exceptions to the Democrats' improved showings in populous areas, but they were rare.
For instance, in relatively urban areas of the Dakotas, Trump actually fared better than Romney did. North Dakota's most populous counties -- Cass (Fargo), Burleigh (Bismarck), Ward (Minot), and Grand Forks -- all shifted their margins towards Trump by between seven and 16 percentage points. And in South Dakota, each of the five most populous counties -- Minnehaha County (Sioux Falls), Pennington County (Rapid City), Lincoln County (the southern suburbs of Sioux Falls), Brown County (Aberdeen), and Brookings County (Brookings) -- shifted toward Trump by between four and 19 percentage points.
Wealthy Resort Areas
Clinton improved her standing in wealthy enclaves of three of the country's most solidly Republican states.
In Utah's Summit County (Park City), Romney, a Utah native, won by about five points, but Clinton won by 14. In Idaho's Blaine County (Sun Valley), Obama won by 20 points but Clinton won by 29. And in Wyoming, Teton County (Jackson) gave Clinton a winning margin 15 points bigger than Obama's in 2012.
The Quirkiest Results
The oddest pattern we noticed involved voter results that didn't include either Trump or Clinton.
In Vermont, Clinton won 57 percent of the vote -- well below Obama's 67 percent four years earlier. But most of the vote share bled by Clinton didn't go to Trump -- he actually won a slightly smaller percentage of the vote statewide than Romney did in 2012.
Some went to third-party candidates: 3 percent for Libertarian Gary Johnson and 2 percent for Green Party nominee Jill Stein. A stunning 7 percent of the statewide tally went to write-in votes, including nearly 6 percent for favorite son Bernie Sanders, who had lost the Democratic primary to Clinton. That was up significantly from 2012, when write-in votes accounted for less than half a percent of presidential ballots cast.
Another 1.4 percent of Vermonters' presidential ballot lines in 2016 were left blank.
*CORRECTION: A previous version of this mistakenly referred to Plainfield, Conn., as a county.