Improvements in Republican and Democratic vote shares in recent elections by state

These are the states in which presidential nominees most strongly raised their party's vote shares.

-If the nominee raised his party's national vote share, they are the states in which he raised his party's vote share by at least 1% more than he did nationally.

-If the nominee saw his or her party's vote share decline nationally, they are the states in which he or she raised his or her party's vote share at all.

States (and DC) are coloured in the colour of the party that carried them in the election in question. Bolded states are states the party flipped. Italicised states (which are shaded more lightly on maps) are states in which the other party raised its vote share more in the same election. 

Benjamin Harrison, 1892

Delaware +5.04%
South Carolina +1.76%
New Hampshire +0.77%

William Howard Taft, 1912

(none)

Herbert Hoover, 1932

(none)

Gerald Ford, 1976 (national increase: -12.65%)

(none)

Ronald Reagan, 1980 (national increase: +2.73%) (map)

Arkansas +13.20%
North Dakota +12.57%
Nevada +12.37%
South Dakota +11.62%
Oklahoma +10.54%
Utah +10.33%
Florida +8.88%
Georgia +7.99%
Texas +7.31%
Idaho +6.58%
South Carolina +6.44%
Nebraska +6.34%
Alabama +6.14%
Tennessee +5.76%
Kansas +5.36%
Louisiana +5.25%
North Carolina +5.08%
Arizona +4.24%
New Mexico +4.22%
Montana +3.98%
Virginia +3.74%

Ronald Reagan, 1984 (national increase: +8.02%) (map)

Georgia +19.22%
Maine +15.22%
Rhode Island +14.46%
South Carolina +13.98%
Vermont +13.55%
North Carolina +12.60%
Connecticut +12.57%
Delaware +12.57%
Mississippi +12.43%
Arkansas +12.34%
Alaska +12.30%
Hawaii +12.20%
Alabama +11.79%
Kentucky +10.97%
New Hampshire +10.92%
Michigan +10.24%
West Virginia +9.81%
Florida +9.80%
Louisiana +9.57%
Massachusetts +9.32%
Virginia +9.26%
Tennessee +9.14%

Reagan's most powerful gains were very regionally concentrated in both of his elections: in the South and Mountain/Plains West in 1980, and in the South and New England (where he seems to have inherited most of the Anderson vote) in 1984.

George H. W. Bush, 1988 (national increase: -5.40%) (map)

District of Columbia +0.57%
Tennessee +0.05%

This was the third election in a row in which the Republican vote share rose in Tennessee.

George H. W. Bush, 1992 (national increase: -15.92%)

none (last time one of the two major parties failed to increase its vote share in any state)

Bob Dole, 1996 (national increase: +3.26%) (map)

Kansas +15.41%
Alaska +11.34%
Utah +11.01%
Idaho +10.15%
Wyoming +10.11%
Colorado +9.93%
Montana +8.99%
Texas +8.20%
Nevada +8.18%
Missouri +7.32%
Nebraska +7.07%
Oregon +6.53%
South Dakota +5.83%
Arizona +5.82%
Oklahoma +5.61%
California +5.60%
Washington +5.33%
North Carolina +5.29%
New Mexico +4.52% 

Dole gained almost across the board west of Beatson's Line, in most places (although not everywhere) more than Bill Clinton. He improved scarce few places east of it, with one geographic outlier being North Carolina, the site of the only county casting over 100,000 votes that he flipped outside California (Wake, NC--although Clinton also flipped a major county in the state in 1996, Mecklenburg). 

Outside Missouri, there was little intimation of coming Republican breakthroughs. Eight states belong to at least one of the two groups of 'Greater Appalachia' or the states that voted twice for Bill Clinton and then never voted Democratic again; these eight also include the four states in which the Democratic vote share declined for the five straight elections from 2000 through 2016, the seven states in which, as of 2021, Al Gore's vote share marks the high-water point for the Democracy in the 21st century, and the three states in which either Bill Clinton or Jimmy Carter holds (again, as of 2021) the record for most popular votes won by a Democrat for president. This general area has increasingly become the centre of the Republican Party this century (with its largest state, Tennessee, becoming the Republicans' best raw vote margin state in 2020). Bob Dole improved by over 1% more than his national improvement in Missouri and Oklahoma (although in Oklahoma, Clinton's improvement outpaced his own). Otherwise, however, he improved by no more than 3.54% (in Kentucky) in any of the others. Only in Missouri, Kentucky, and Oklahoma did his improvement outpace his national improvement at all. In Louisiana, his vote share was actually slightly lower than George H. W. Bush's. 

Likewise, in the Upper Midwest states that Bush surprisingly put into play in 2000, Dole improved on HW Bush by 1.70% in Wisconsin, 2.65% in Iowa, and 3.11% in Minnesota.

George W. Bush, 2000 (national increase: +7.16%) (map)

Wyoming +17.95%
West Virginia +15.16%
Idaho +14.99%
Arkansas +14.51%
Montana +14.33%
South Dakota +13.81%
North Dakota +13.72%
Maine +13.21%
Louisiana +12.61%
Utah +12.46%
Oklahoma +12.05%
Kentucky +11.62%
Minnesota +10.54%
Texas +10.54%
Vermont +9.61%
Indiana +9.52%
Missouri +9.18%
Wisconsin +9.13%
Ohio +8.95%
New Hampshire +8.70%
Nebraska +8.60%
Mississippi +8.41%
Iowa +8.30%

Interestingly, in the very same election in which Wyoming became the Most Republican State for the first time, it was also the state in which Bush increased the Republican vote share the most--by over 2.75%. 

In the eight combined 'Greater Appalachia'/'twice-Clinton-thereafter-red' states (which we could call 'Greater Appalachia + 2'), Bush improved by over 1% more than his national improvement in six (all save Tennessee and Alabama). His second-biggest improvement was in West Virginia, a traditionally Democratic state that Bush not only carried but carried fairly comfortably. He also improved by over 1% more than his national improvement in the three Upper Midwest Dukakis states of Wisconsin, Iowa, and Minnesota (although not in the more industrial states of Pennsylvania and Michigan), most of the Plains West (save Dole's home state of Kansas), and much of the interior Mountain West (the exceptions would turn out to be the four states in the region that voted for Biden in 2020). Interestingly, he also strongly improved throughout the Yankee Kingdom, including future bluest state in the country Vermont. Finally, even though it was seen as surprising on election night that he was having such difficulty carrying it, he improved by more than 1% more than his national improvement in Ohio.

Minnesota, Missouri, Wisconsin, and Iowa (and, for that matter, Ohio) were states in which Reagan never improved by more than 1% over his national improvement in either of his elections. Reagan's gains in these states in 1984 were 6.98%, 8.86%, 6.29%, 1.96%, and 7.39%, respectively, meaning that he never improved the Republican vote share in any of them by as much as Bush did in 2000.

George W. Bush, 2004 (national increase: +2.86%) (map)

Hawaii +7.80%
Rhode Island +6.76%
Alabama +5.98%
New Jersey +5.95%
Tennessee +5.65%
Connecticut +5.51%
Oklahoma +5.26%
New York +4.85%
Utah +4.71%
Massachusetts +4.28%
Louisiana +4.17%
West Virginia +4.14%
Kansas +3.96%

Five of these states are 'Greater Appalachia +2' states (including both in which he did not improve by over 1% more than his national improvement in 2000--Gore's home state of Tennessee, and Alabama). Bush's surge in the rural Upper Midwest seems to have stalled out in 2004, though, although he inched forward enough to flip Iowa. In terms of the national popular vote, aside from Greater Appalachia + 2, the main driver behind Bush's improvement seems to have been the Northeast and, more specifically, '9/11 belt voters' (as three of the five Northeastern states in which he improved by at least 3.86% were New York, Connecticut, and New Jersey). He also significantly improved in Rhode Island--where Trump would do unusually well in 2016--and, less explicably, Massachusetts. Massachusetts had an unusually large third-party vote in 2000, but Bush improved by substantially more than John Kerry (the state's senator) improved on Gore in it. (Boston was where two of the hijacked planes in 2001 departed from.)

John McCain, 2008 (national increase: -5.08%) (map) 

Arkansas +4.41%
Louisiana +1.84%
Tennessee +0.10%
Oklahoma +0.08%

McCain oversaw a decline in the Republican presidential vote share fairly similar to that in 1988. But he improved the Republican vote share in four states, and in one of them, fairly significantly. These were all 'Greater Appalachia + 2' states. Except for Oklahoma, all of these states had Kerry-McCain counties. (In Oklahoma, Bush had carried every county in 2004, but Gore had carried some counties in 2000, none of which Obama reclaimed in 2008.)

Three of these states are three of the four states in which the Democratic vote share declined in every election from 2000 through 2016; in the fourth, West Virginia, both parties' vote shares declined in 2008.

Mitt Romney, 2012 (national increase: +1.55%) (map) 

Utah +10.21%
West Virginia +6.59%
Montana +5.84%
Indiana +5.22%
North Dakota +5.07%
South Dakota +4.73%
Missouri +4.33%
Illinois +3.95%
Wyoming +3.86%
Michigan +3.75%
Wisconsin +3.58%
Nebraska +3.27%
Kansas +3.10%
Kentucky +3.09%
Delaware +3.03%
Nevada +3.03%
Idaho +3.01%
Tennessee +2.58%

Early on election night 2016, Nate Silver wrote,

The differences between national polls, which often showed a very tight race for the popular vote, and polls of swing states, where President Obama usually maintained an advantage, were a source of intrigue this year. 

It could be that Mitt Romney’s performance in strongly red-leaning states, which were sparsely polled this year, accounts for much of the difference, allowing him to rack up votes without helping himself in the Electoral College.

Mr. Romney currently leads by 27 points in Tennessee, by 22 points in Kentucky, by 16 points in South Carolina and by almost 40 points in Oklahoma.

Those leads ended up getting whittled down to 20.40%, 22.69%, 10.47%, and 33.54%, respectively; and in South Carolina and Oklahoma, he didn't even manage a > 2% increase over McCain (he increased over McCain by 0.69% and 1.12% in the two states, respectively).

However, Romney's strongest increases did end up being essentially electorally useless (if one considers that, despite having voted for Obama in 2008, Indiana was essentially factored into Romney's map as a given, along with the McCain states). Romney's two strongest increases were in states that had voted for McCain in 2008 by double digits. His next five strongest increases were in five historically red states in which Obama had cut the margin to under 10% in 2008 (coming very close in Missouri and Montana, and of course, narrowly carrying Indiana outright). His strongest improvement in a swing state was 3.75%, in Michigan. The only swing states in which Romney improved by more than 1% more than he did nationally were Michigan, Wisconsin, and Nevada, none of which he came particularly close to carrying. In the one swing state he did flip--North Carolina--he improved on McCain by just 1.01%. 

Donald Trump, 2016 (national increase: -1.11%) (map)

West Virginia +6.20%
Iowa +4.97%
Maine +4.89%
North Dakota +4.64%
Ohio +4.00%
Rhode Island +3.66%
South Dakota +3.64%
Missouri +3.01%
Michigan +2.79%
Indiana +2.69%
Mississippi +2.65%
Hawaii +2.19%
Kentucky +2.03%
Delaware +1.74%
Pennsylvania +1.59%
Alabama +1.53%
New York +1.35%
Wisconsin +1.33%
Tennessee +1.24%
Montana +0.82%
New Jersey +0.76%
New Hampshire +0.58%
South Carolina +0.38%
Louisiana +0.31%
Connecticut +0.20%
Arkansas +0.005%

The map of states in which Trump improved on Romney's vote share is almost a reverse of the map of states in which Dole improved on George H. W. Bush's vote share by more than 1% more than his national improvement in 1996. Trump improved in most of the states east of Beatson's Line, with the interesting exceptions of most of the Atlantic South (deceptively, as it turned out, including Florida), Illinois, Massachusetts, Vermont, and--perhaps forebodingly--Minnesota. West of Beatson's Line (on the North American continent), Trump improved only in the Dakotas and Montana. (He also improved in the Polynesian state of Hawaii.) He improved on Romney in every state he flipped save Florida.

Donald Trump, 2020 (national increase: +0.77%) (map)

Utah +12.59%
Idaho +4.58%
Hawaii +4.24%
New Mexico +3.42%
California +2.70%
Florida +2.20%
Nevada +2.17%
North Dakota +2.15%
Iowa +1.94%
Washington +1.94%
Arkansas +1.83%
Illinois +1.79%

This is a somewhat disjointed map of strong-improvement states. There also aren't very many states (although California, Florida, and Illinois are amongst the six most populous states). The only 'Greater Appalachia + 2' state is Arkansas, the state where Hillary Clinton was First Lady for most of the 1980s. In Tennessee, the Republican vote share actually declined in 2020, for the first time since 1996. But for Arkansas, Trump would have been the first Republican nominee since 1992 not to improve more than he did nationally in any of the 'Greater Appalachia + 2' states. 

In Kentucky and Alabama, Trump's vote share likewise slightly declined, the first time the Republican vote share had done this in either of those states in the 21st century save in 2008. In West Virginia, Trump inched forward by 0.12%, by far the smallest increase in the GOP vote share this century apart from 2008, when it declined. In Louisiana, Oklahoma, and Missouri, he likewise made tiny, incremental improvements of 0.36%, 0.05%, and 0.03%, respectively. (At the same time, in none of these states--in fact, in no state full stop--did he reach 71.54% or 72.79%--Bush's and Romney's vote shares in Utah in 2004 and 2012, respectively. In Missouri, he remained below 60%, as well as below Bush's and Romney's vote shares in the behemoth of Texas in 2000, 2004, and 2012, or Bush's vote share in similarly-sized Indiana in 2004, or even Trump's own 2020 vote share in similarly-sized Indiana--or the Republican vote share in similarly-sized Tennessee in any of 2008, 2012, 2016, or 2020. In other words, while one might not expect the Republican vote share in these states to simply increase indefinitely, there is no particular reason why this election, in particular, should have seen the Republican vote share top out.) In Louisiana, McCain's vote share remains the Republican high-water mark this century; in Oklahoma, Romney's does.

There is little overlap with the map of states in which Trump improved in 2016. Only Hawaii, North Dakota, Iowa, and Arkansas are on both maps; and Biden improved more than Trump in North Dakota (and even Iowa). Iowa is the only state in which Trump surprised analysts in 2016, in which he improved by over 1% more than he improved nationally in 2020. He also had non-trivial improvements in Ohio (1.58%) and Wisconsin (1.60%), although these were much less than, say, Bush's improvements in West Virginia and Louisiana in 2004. In Michigan, Minnesota, and Pennsylvania, he improved by less than 1% (and by less than his national improvement)--by 0.34%, 0.36%, and 0.66%, respectively. In Maine, he actually declined, by 0.85%. 

There is also a lack of continuity with earlier Republican gains in the 21st century. Only Arkansas was amongst the four states in which McCain improved on Bush's '04 vote share in 2008. Only Hawaii and Utah were amongst the states in which Bush improved by more than 1% more than his national improvement in 2004. Only Utah, Idaho, Arkansas, North Dakota, and Iowa were amongst the states in which Bush improved on Dole by more than 1% more than he did nationally in 2000. That may seem like a lot of states, but in 2016, Trump improved on Romney in 15 of these states (whilst getting a lower national vote share than Romney). (As for how many consecutive elections one could expect a party's vote share to rise in a given state, the GOP's vote share rose in Tennessee for six straight elections from 1996 through 2016.)

In addition, outside Utah--where Evan McMullin had gotten over 20% (or 1/5) of the vote in 2016--Trump's biggest improvement in any state was smaller than the best Republican improvement in any of the elections from 1996 through 2016 save in 2008 (and it wasn't much better than McCain's improvement in Arkansas in 2008). In fact, if one also sets aside Idaho (where Biden improved more than Trump), then Trump's best improvement was smaller than the best Republican improvement in any of the elections from 1996 through 2016 full stop.

All that said, Donald Trump was the first incumbent president to fail in his bid for an additional consecutive term despite improving his party's vote share in at least one state since 1892. Taft in 1912, Hoover in 1932, Ford in 1976, Carter in 1980, and George H. W. Bush in 1992 all failed to improve their party's presidential vote share in a single state in those elections. Harrison improved the Republican vote share in three states in 1892 (Delaware, South Carolina, and New Hampshire); the above-listed twelve states are, of course, not the only states in which Trump raised the Republican vote share in 2020 (just the ones where he did so by more than 1.77%). Trump increased his vote share by any percentage in 33 states (and the District of Columbia).

---

Jimmy Carter, 1980 (national increase: -9.07%)

none (last time the Democratic vote share did not increase in any state)

Walter Mondale, 1984 (national increase: -0.45%) (map)

District of Columbia +10.49%
North Dakota +7.54%
Iowa +7.29%
Massachusetts +6.68%
Montana +5.75%
Washington +5.54%
California +5.36%
Nevada +5.08%
Oregon +5.07%
South Dakota +4.84%
Arizona +4.30%
Utah +4.11%
Colorado +4.05%
Pennsylvania +3.51%
Alaska +3.46%
Minnesota +3.22%
Nebraska +2.77%
New Hampshire +2.60%
New Mexico +2.45%
Vermont +2.40%
New York +1.84%
Wisconsin +1.84%
Illinois +1.58%
Idaho +1.20%
New Jersey +0.64%
Rhode Island +0.35%
Connecticut +0.31%
Wyoming +0.27%

Michael Dukakis, 1988 (national increase: +5.09%) (map)

Oklahoma +10.61%
Hawaii +10.45%
Nebraska +10.31%
Colorado +10.16%
South Dakota +9.98%
Kansas +9.96%
Wyoming +9.77%
Idaho +9.62%
North Dakota +9.17%
Iowa +8.82%
Connecticut +8.04%
Montana +8.02%
Missouri +7.87%
New Mexico +7.67%
Rhode Island +7.62%
West Virginia +7.60%
Oregon +7.54%
Utah +7.37%
Texas +7.24%
Washington +7.19%
Vermont +6.77%
Alaska +6.40%
Wisconsin +6.39%
California +6.29%
Arizona +6.20%

There is actually a significant degree of overlap between these states, and the states in which Dole improved on HW Bush by more than 1% more than his national improvement. Dukakis improved strongly in most of the states west of Beatson's Line, and east of it, he improved strongly in Missouri (which Dole also improved strongly in). Dukakis also improved strongly in the Upper Midwest states of Wisconsin and Iowa. Otherwise, strong improvements east of Beatson's Line were in Vermont, Connecticut, and West Virginia. In most of the South east of Beatson's Line, Dukakis improved by less than he did nationally, in many cases significantly so. (In most Southern states east of Beatson's Line--but in neither Oklahoma nor Texas--there was also at least one county in which HW Bush improved on Reagan's vote share.) Oklahoma--somewhat surprisingly, the site of Dukakis' strongest improvement--and Texas behaved more like the West than like the South in this election. (Dukakis also had a relatively strong 5.88% improvement in Louisiana.)

From Dukakis' particularly strong improvements, one might also have imagined that the Democratic nominee in 1992--especially if a Southerner--would be likelier to carry Texas than Florida. (Clinton carried neither in 1992, but he came closer in Florida, and he flipped Florida in 1996, whereas he never carried Texas.)

Bill Clinton, 1992 (national increase: -2.64%) (map)

Arkansas +11.02%
Tennessee +5.53%
Georgia +3.97%
New Hampshire +2.58%
Florida +2.38%
South Carolina +2.30%
District of Columbia +1.99%
Mississippi +1.70%
Maryland +1.60%
Louisiana +1.52%
Virginia +1.36%
Alabama +1.02%
North Carolina +0.94%
Kentucky +0.67%
New Jersey +0.35%
Delaware +0.04% 

In an election in which Clinton's actual vote share was lower than Dukakis', Clinton increased his vote share almost across the board in the South east of Beatson's Line (but not in Texas or Oklahoma--nor in Missouri or West Virginia, the two Southern states east of Beatson's Line in which Dukakis had improved his vote share by more than 1% more than he did nationally). Outside the South, Clinton improved on Dukakis' vote share only in two Northeastern states, New Hampshire and New Jersey. These had been HW Bush's two best states in the Northeast in 1988, and Clinton carried them only narrowly in 1992.

Bill Clinton, 1996 (national increase: +6.23%) (map)

Massachusetts +13.93%
Maine +12.85%
Rhode Island +12.67%
New Jersey +10.77%
Connecticut +10.62%
New Hampshire +10.41%
Arizona +10.00%
New York +9.74%
Florida +9.02%
Hawaii +8.84%
Utah +8.65%
Delaware +8.28%
North Dakota +7.95%
Michigan +7.92%
Wisconsin +7.68%
Minnesota +7.62%
Vermont +7.24%

Clinton's strongest gains in 1996 were in the Northeast and the Upper Midwest (with, in this case, North Dakota behaving more like an Upper Midwest state than like a Western state). The two states he flipped--Arizona and Florida--were both geographic outliers. But for them, Clinton could have improved his national vote substantially without helping himself at all in the Electoral College (although he came within seven points in North Dakota in 1996). 

Al Gore, 2000 (national increase: -0.86%) (map)

Delaware +3.16%
Connecticut +3.08%
New Jersey +2.41%
California +2.35%
Maryland +2.32%
Nevada +2.05%
Pennsylvania +1.43%
Rhode Island +1.28%
Kansas +1.16%
Florida +0.82%
New York +0.74%
Washington +0.32%
Illinois +0.28%

Interestingly, the only state whose margin actually swung more Democratic in 2000 was Maryland--also the only state in which the Republican vote share has declined for four elections in a row (from 2008 through 2020). It was also home to one of the only two Dole-Gore counties (although it was the smaller of the two), Charles County, Maryland. 

John Kerry, 2004 (national increase: -0.11%) (map)

Vermont +8.31%
Alaska +7.85%
Montana +5.20%
Colorado +4.63%
Maine +4.48%
Oregon +4.39%
District of Columbia +4.02%
New Hampshire +3.44%
Minnesota +3.18%
Washington +2.66%
Idaho +2.62%
North Dakota +2.44%
Ohio +2.25%
Massachusetts +2.14%
Nevada +1.90%
Wisconsin +1.87%
Wyoming +1.37%
New Mexico +1.14%
Virginia +1.04%
South Dakota +0.88%
California +0.86%
Iowa +0.69%
North Carolina +0.38%
Pennsylvania +0.32%
Texas +0.24%
Illinois +0.22%
South Carolina +0.0009%

Kerry's national vote share barely declined relative to Gore's, so there are a lot of states on this map, many of them in light blue (meaning Bush improved in them more than Kerry improved on Gore). Still, it wasn't as though the default was for Kerry to improve on Gore, as there were many states in which he did not do so. Kerry did not improve on Gore in many places in the South, so it is noteworthy where he did: Virginia, North Carolina, South Carolina (infinitesimally), and Texas. Other than South Carolina, these were all home to some of the largest counties Kerry flipped nationally: Fairfax, Mecklenburg, Guilford, and Travis. 

Kerry also improved strongly in much of the Mountain West and Upper Midwest (including the Dakotas), as well as in the Yankee Kingdom. His improvements in the Upper Midwest do not seem to have been durable--in Wisconsin, Iowa, Minnesota, and the Dakotas, the only county he flipped in 2004 that has gone on to vote Democratic in 2020 is Cook, MN. In the Mountain West, however, most of his improvements seem to have been driven by 'ski resort and mountain nirvanas'; amongst the (decently-sized) counties he flipped in these states were La Plata, Gunnison, Eagle, and Routt, CO; Missoula, MT; and Teton, WY, all of which have gone on to vote Democratic in every subsequent election. In Idaho and New Mexico, he didn't flip anything, but he expanded Gore's margin in Blaine County, Idaho by 16.5%, and in Taos County, NM by 10.0%. (Michael Barone noted Taos County--along with larger Santa Fe County, where Kerry expanded Gore's margin by 6.8%--as two counties 'thick with trust-funder liberals' that voted for Obama in the 2008 primary.)

Ohio was an odd outlier; Kerry's vote share fell backward in every state bordering Ohio save Pennsylvania, where it went up less than Bush's. Ohio was also home to the only county casting over 100,000 votes, that flipped from Bush in 2000 to Kerry in 2004, that has subsequently voted Republican (and indeed voted for Trump twice): Stark County, Ohio.

Barack Obama, 2008 (national increase: +4.66%) (map)

Hawaii +17.84%
Indiana +10.69%
North Dakota +9.12%
Nebraska +8.92%
Montana +8.69%
Delaware +8.59%
Vermont +8.52%
Utah +8.41%
New Mexico +7.86%
Nevada +7.27%
Virginia +7.15%
Illinois +7.10%
California +6.70%
Colorado +6.64%
Wisconsin +6.52%
South Dakota +6.31%
Connecticut +6.28%
Michigan +6.20%
North Carolina +6.12%
Maryland +6.01%
Idaho +5.83%

Obama improved by over 1% more than his national improvement in most of the interior Mountain West (exceptions being McCain's home state of Arizona, as well as Wyoming); in most of the Midwestern states adjoining the Great Lakes (one notable exception being Ohio); in the free-soil Plains West save Kansas; and in the Atlantic Upper South. Rounding out the list were Vermont, Connecticut, Hawaii, and the giant state of California.

Aside from Ohio, notably absent were Pennsylvania, Minnesota, Iowa, any Atlantic South states south of North Carolina (including Florida), and the Pacific Northwest. There also aren't many states in the Northeast. Vermont often follows offbeat patterns, but Connecticut is hard to explain. None of the 'Greater Appalachia + 2' states are here, including Missouri (which Obama made a play for and nearly carried, but where Bush had done decidedly less well in 2004 than he had done in Indiana--which Obama did carry, with a massive swing of over 20%).

Barack Obama, 2012 (national increase: -1.87%) (map)

Alaska +2.92%
New Jersey +1.11%
Mississippi +0.79%
Louisiana +0.65%
New York +0.47%
Maryland +0.05%

Other than Maryland and Alaska, these states aren't really part of a long-term pattern of Democratic improvements, as far as I can tell. 

Hillary Clinton, 2016 (national increase: -2.88%) (map) 

Utah +2.71%
Texas +1.86%
California +1.49%
Georgia +0.16%

These four states were, collectively, home to six of the 16 Romney-Hillary counties (not counting Kenedy, TX). Orange, CA was the largest one, and others that cast over 100,000 votes in these states were Fort Bend, TX; Salt Lake, UT; and Cobb and Gwinnett, GA. (And, although Hillary Clinton won Salt Lake County with a low plurality in a three-way race, she also improved on Obama's 2012 vote share by 3.44% in the county.) Other Romney-Hillary counties in these states were Nevada, CA, Henry, GA, and Summit, UT, none of them trivially-sized; the smallest, Summit, cast about 1.82% of Utah's total vote in 2016, and unlike in the case of Salt Lake County, Hillary Clinton won it with an absolute majority.

Joe Biden, 2020 (national increase: +3.13%) (map)

Utah +10.19%
Vermont +9.41%
Colorado +7.24%
Oregon +6.38%
Alaska +6.22%
New Mexico +6.03%
Minnesota +5.96%
New Hampshire +5.73%
Delaware +5.65%
Massachusetts +5.59%
Idaho +5.58%
Kansas +5.51%
Nebraska +5.47%
Washington +5.43%
Maine +5.26%
Maryland +5.03%
Rhode Island +4.98%
Montana +4.80%
Arizona +4.78%
Connecticut +4.69%
Wyoming +4.67%
North Dakota +4.53%
Virginia +4.38%

It's interesting to compare this map with the one in 2008.

In 2008, Obama made major gains in the traditionally Republican states of Virginia, North Carolina, and Colorado. In 2020, Biden made major gains in Virginia and Colorado, the first time the Democrats had improved by more than 1% more than their national gain in these states since 2008. Not, however, in North Carolina (where Biden improved by 2.42%). Once again, Biden made a significant gain in Maryland--the fourth time in the last six elections in which the Democrat had either improved by more than 1% more than his or her national improvement, or had improved while declining nationally. 

Obama had also made a significant gain in Nevada, but Nevada was actually one of a handful of states to swing more Republican in 2020.

California--one of the few states in which Hillary Clinton had improved on Obama '12's vote share--was another of the few states to swing more Republican in 2020.

New England acted as a unified bloc, uniformly giving Biden a particularly strong boost. However, New England is noticeable in this respect because, outside New England, Biden didn't increase by all that much in the rest of the Northeast. Nor did he increase by that much in the Midwest (unlike Obama in 2008)--with the noteworthy exception of Minnesota, a state Obama had not improved significantly on Kerry in in 2008, and which the Trump campaign had targeted in 2020 as a possible turnover state. 

Outside the DC-influenced states of Virginia and Maryland, Biden's gains weren't actually that disproportionately great in the Atlantic South, even in his breakthrough state of Georgia (where he improved on Hillary Clinton by 3.83%).

In fact, much like Romney in 2012, Biden's gains were fairly inefficiently distributed. Only one of the states he flipped was one in which he made a gain that was more than 1% more than his national gain--Arizona. In Michigan, Biden improved on Hillary by 3.35%; in Pennsylvania, by 2.55%; and in Wisconsin, by 3.00%. In Pennsylvania and Wisconsin, he improved by less than he did nationally. In states that were thought to be potentially competitive but which Biden did not flip, he improved by 3.15% in Iowa, by 1.68% in Ohio, and by 3.24% in Texas. His gain was particularly small in Ohio. 

One commonality between 2008 and 2020 is the preponderance of strong gains in the Mountain and free-soil Plains West.

Of course, I have basically been defining a 'strong gain' in terms relative to a nominee's overall gain. Even Biden's increase over Hillary Clinton in Ohio would--if translated nationally--have been enough to get to 49.86% nationally, which would almost certainly secure a national popular vote win. It's just that Biden's gains in Ohio (and, to a lesser extent, other Frost Belt/industrial states, including ones he flipped) played a relatively insignificant role in the boosting of Biden's overall vote share. But it shouldn't really be that comforting to the Republican Party that Biden improved on Hillary Clinton in Ohio by 'only' 1.68%, or in Pennsylvania by 'only' 2.55%. 

(Conversely, Trump's 1.94% gain in Iowa is impressive in the context of his national improvement, but it is smaller than Biden's gain in any of Pennsylvania, Wisconsin, or Michigan, and not much more than Biden's gain in Ohio. If translated nationally, it would have gotten Trump to just 48.03%, which is usually not enough to win the national popular vote in a two-party contest.)

Biden increased the Democratic vote share by some percentage in all 50 states (and the District of Columbia). This is very unusual, and no nominee who has increased his party's vote share in every state has ever lost. These are the nominees who have increased their party's vote share in every state:

Warren Harding, 1920 (national increase: +14.20%)
Franklin Roosevelt, 1932 (national increase: +16.61%)
Dwight Eisenhower, 1952 (national increase: +10.11%)
Richard Nixon, 1972 (national increase: +17.25%)
Jimmy Carter, 1976 (national increase: +12.56%)
Ronald Reagan, 1984 (national increase: +8.02%)
Bill Clinton, 1996 (national increase: +6.23%)
George W. Bush, 2000 (national increase: +7.16%)*
Joe Biden, 2020 (national increase: +3.13%)

* George W. Bush increased the Republican vote share in all 50 states in 2000, but not in the District of Columbia.

This seems to be getting more common, more recently; and the national increase required to lead to an improvement in every state seems to be getting smaller. It didn't happen at all in the 1800s, and indeed, from 1920 through 1984, the occurrence of this phenomenon could--almost--be used as an indicator of a landslide election. From 1920 through 1984, every election in which this happened was either a landslide (1920, 1932, 1952, 1972, 1984), or followed a landslide (1932, 1976). And only in the case of three landslide elections in that period was neither it nor the election following it an election in which one of the two parties increased its vote share in every state. Two--1936 and 1956--were landslides following upon a landslide for the same party. (That doesn't explain why the elections following them were not elections in which the out-party improved its vote share in every state, of course, but the out-party also failed to do this in 1988, despite improving significantly nationally.) 

The third exception is 1964. In this case, the subsequent election was a three-way contest in which the out-party increased its vote share by just 4.77% (but the in-party certainly wasn't winning a second consecutive landslide). Of course, that doesn't explain why the landslide-winning party didn't itself improve in every state in 1964. But the landslide-winning party also failed to do this in 1928.

In any case, it was surprising how many elections were not elections in which one of the two parties failed to improve its vote share in every state. In 1988, Dukakis, improving by 5.09% on Mondale nationally, went backwards in Georgia and Tennessee (and the District of Columbia). In 1960, Kennedy, improving on Stevenson '56 by 7.75%, went backwards in Alabama, Arkansas, Georgia, Mississippi, Oklahoma, and Tennessee. (Tennessee was also one of the states in which the Democratic vote share declined in 2008, making it perhaps 'best at bucking the national trend in generic Democratic years'.) In 1940, Wendell Willkie, improving on Landon nationally by 8.24%, went backward in Maine, New Hampshire, North Carolina, and Vermont.

In the case of 1916, it might be surprising that both nominees failed to improve in every state, but particularly in the case of Charles Evans Hughes, who improved on William Howard Taft nationally by 22.95%. Hughes underperformed Taft's '12 vote share in Oklahoma. (Wilson underperformed his 1912 vote share in Florida and North Carolina.)

Now, until 1984, no nominee improved his party's vote share in every state without improving his party's national vote share by at least 10% (which Willkie, Kennedy, and Dukakis--and Wilson in 1916--all failed to do). In 1996, Clinton managed to do so with a national increase of just 6.23%. Bush did so (excepting the District of Columbia) with a national increase of just 7.16% in 2000.

However, Biden's doing so with a national increase of just 3.13% in 2020 was a truly abrupt departure. This is only a little over half as large as the next-smallest national increase that translated into an increase in every state (Clinton's 6.23% increase in 1996). It is over 1.5% smaller than Obama's 4.66% national increase in 2008 (amid which Obama failed to improve his party's vote share in Oklahoma, Tennessee, Arkansas, West Virginia, or Massachusetts). It is also, obviously, significantly smaller than a number of national improvements in the 20th (and 19th) centuries that failed to translate into an improvement in every state (such as Tilden's 7.14% increase in 1876--he declined in Kansas; Lincoln's 6.71% increase in 1860--he declined in Massachusetts and Vermont; and the Whigs' collective 11.67% increase in 1836--they declined in Connecticut, Kentucky, Maine, New Hampshire, New York, and Rhode Island).

So this metric might not say all that much about Biden's improvement in 2020 being particularly impressive (after all, one would be hard-pressed to put Biden's national improvement in a category with Harding's in 1920, FDR's in 1932, etc., but apart from Obama's in 2008, Dukakis' in 1988, Kennedy's in 1960, Willkie's in 1940, and Tilden's in 1876). It appears that Biden's national improvement was simply very unusually evenly spread out for its size, which may be interesting--but the small size of his national improvement also means that the fact that he improved in every state means that nowhere was his improvement particularly extraordinary. Whether this is good or not, I'm not sure. It sounds good for a party's gains to be very evenly distributed (although the evenness of the distribution of Biden's improvement is so unprecedented that it seems unnecessary). On the other hand, it didn't hurt Obama in the Electoral College that he failed to improve in Oklahoma, Tennessee, Arkansas, or West Virginia (or Massachusetts). Nor did it help Biden, or even help his party's prospects in the Senate, that he improved his party's vote share marginally in Oklahoma, Tennessee, Arkansas, and West Virginia (or that he managed to improve on Hillary Clinton in Florida--his closest near-miss--by 0.04%). (This is even in a long-term sense--while states such as Utah and Alaska may be trending more purple in the long run, Oklahoma, Tennessee, Arkansas, and West Virginia will remain solidly red until a realignment takes place that renders the current map unrecognisable.)

To take it to an extreme, it probably wouldn't be a good thing (for the Republican Party) if Trump--while improving his national vote share by 0.77%--had managed to improve his vote share in all 50 states. (As it is, he improved his vote share in nearly two-thirds of the states, and accordingly, outside Utah and, perhaps, Hawaii, the degree of his increases was underwhelming everywhere, compared to other recent elections--including in states he failed to hang onto, such as Pennsylvania, Wisconsin, and Michigan.)