Some observations

on this post (which it is hard and slow to add text to now).

-Today, there are five states in which all eight of the top vote shares belong to Republicans. As of October 1996 or October 2000, there were three. The original three were Kansas, Nebraska, and Vermont--three of the most historically Republican states. The five today are those three plus Wyoming and Indiana. 

Interestingly, it appears that neither Wyoming nor Indiana was a state in which Republicans had ever held all top eight vote shares before the 21st century. (These are, I believe, the states in which it had ever been the case at some point before 2000 that all eight of the top vote shares belonged to Republicans.) Comparing Indiana and Ohio, in particular, to the rest of the Midwest, we see that, in the past, Ohio and Indiana had tended to give relatively low vote shares to everybody. As late as October 1952, the eighth-highest vote share ever earned in Indiana was Theodore Roosevelt's 53.99%; and the eighth-highest vote share ever earned in Ohio was Grant's 54.00% in 1868. They were also, as noted in the parent post, frequently won with pluralities. 

The rest of the free-soil Midwest, except for Illinois (Wisconsin, Iowa, Minnesota, and Michigan) were all states in which, at some point in the distant past, Republicans held all eight of the top vote shares ever earned--at the same time as all those vote shares were high. In Iowa, for example, Republicans held the top eight vote shares from (at least) December 1908-October 1964, by the end of this period, the lowest of those eight was Eisenhower's 59.06% in 1956. Even as of October 1920, the lowest was Taft's 55.62% in 1908--winning by better than 1.25:1. In Michigan, Republicans held the top eight vote shares from (at least) December 1928-October 1964; the lowest of the top eight was Lincoln's 57.23% in 1860 during this time. In Minnesota, Republicans held the top eight vote shares from (at least) December 1888-October 1932; after December 1908, the lowest of the top eight was either Lincoln's 59.06% in 1864 or Taft's 59.11% in 1908. And in Wisconsin, Republicans held the top eight vote shares from December 1904-October 1932; after December 1920, the lowest of these eight was Frémont's 55.30% (which is above 55%, if not above 55.5%--all the other seven were above 55.5%).

Like its fellow free-soil 'mean centre' states, Illinois never quite made it to a place where all eight top vote shares in it belonged to Republicans. But it came closer than Indiana or Ohio; and, unlike in Indiana and Ohio, it did get to a place where almost all eight of its top vote shares were above 55.5% in the distant past. Between December 1928 and October 1932, the top seven vote shares ever earned in Illinois all belonged to Republicans, and all seven were above 55.5%.

In 1978, Bill Schneider wrote that 

Liberalism is strongest in three specific areas of the North: (1) New England and the Middle Atlantic states above the Mason-Dixon line, (2) the northern tier of the Midwest from Michigan to North Dakota, and (3) the Pacific states, particularly Oregon. What the Northeast, the Upper Midwest, and the Pacific Northwest have in common is a particular cultural heritage; they are the areas of Yankee settlement and influence which share the peculiarly moralistic political heritage of New England Protestantism. 

The regions Schneider describes encompasses all the states in which Republicans had ever held all the top eight vote shares prior to 2000 (if we include 'Nebransas' as part of greater Yankeedom)--but not Indiana or Wyoming. In theory, it also includes both Pennsylvania and New York, but while Republicans have held all of the top eight vote shares in Pennsylvania, they never have done in New York. (And, like in the 'northern tier' states, the lowest of the top eight vote shares was high for at least part of the period when Republicans held all eight of the top vote shares in the state--Taft's 58.84% in 1908, from December 1928-October 1964. New York, on the other hand, was similar to Indiana and Ohio in the sense that it took longer than in other states for all eight of its top vote shares to break 55%.) Schneider notes that 'New York is usually thought of as more Yankee-dominated in its upstate area than Pennsylvania, which has a substantial non-Yankee rural population of Germans, Scots-Irish, and Appalachian whites' (which might explain Pennsylvania's relatively late retention, for a Northern state, of 'never-Republican counties'). But (Schneider continues) 'the greater Catholic and immigrant population of New York' contributed to Pennsylvania's being more strongly Republican than New York in the 'System of 1896' (and, arguably, before that), in spite of these things.

And, arguably, Pennsylvania was the most partisan 'leading state' before recent times. Michael Barone wrote of how New York, in its period as the largest state, had (unlike California today) generally 'voted much like the nation as a whole'. But arguably, California, New York, and Texas (until 2016) have all, in recent elections, been voting atypically for states of their size and influence within the United States. Charles Evans Hughes did not set a top-eight vote share in many states, although those that he did tended to be large: New York, Pennsylvania, New Jersey. But in New York, he did so by getting just 51.53% of the vote. In Pennsylvania, he got 54.26%--only slightly worse than Kerry got in California (54.31%).

(As a side note on 'Nebransas', Schneider actually does--using Phillips as a basis--exclude it from 'greater Yankeedom', quoting Phillips as writing that 'Iowa, Wisconsin, and Minnesota were the states of pre-Civil War Yankee settlement. In Kansas and Nebraska, not only were persons of foreign stock less numerous, but the Anglo-Saxon population had a substantial Southern element.' This might be of interest in noting that, leaving aside the somewhat anomalous case of 1864, the only instances on which either Kansas or Nebraska was the Most Republican State were 1928, 1944, 1960, and 1968, two of which saw the Republican running against a Catholic Democratic nominee. That said, I am dubious of just how 'substantial' this was, given Kansas' and Nebraska's historical voting patterns. Doubtless it played a role, and likely explains county-level anomalies such as, say, Sioux County, Nebraska. But Oregon--whereof Schneider quotes Phillips as writing that '[m]ore than any other state in the West, Oregon resembled the East'--was also said by Phillips to have been substantially peopled by 'ranchers and farmers from Texas, Missouri, and the border states'. This arguably affected Oregon's overall political presentation substantially more than any Southern settlement of Nebransas did those states', as Oregon voted all four times--in some cases quite powerfully--for FDR, and furthermore gave only a tepid plurality to Lincoln in 1860, before actually voting for Horatio Seymour eight years later.)

Some of these states became 'all-top-eight-Republican' states relatively recently (or less long ago than the others). This was particularly the case of Connecticut (December 1956-October 1964). In New Jersey, the top eight vote shares were all Republican from December 1924 to October 1936, which began slightly less recently than Michigan, but at every point in this period, the lowest of the top eight vote shares was below 55%. Connecticut and New Jersey, along with New York and Indiana, were the friendliest free-soil states to the Democracy during the Gilded Age; and in fact, New Jersey underwent a fairly dramatic transformation, after the 1896 election, from having five of its eight top vote shares from Dec. 1876-Oct. 1896 being Democratic (and only one being Republican), to having six (including Hughes') being Republican after Dec. 1916. (Nothing quite similar happened in Connecticut, although from Dec. 1876-Oct. 1896, Samuel Tilden's was the eighth-highest vote share ever attained in the state.) These states seemed to become particularly strongly Republican with the period when the Republican Party became associated with the burgeoning postwar suburbs (some of which had already acquired a 'large-suburb' character before the war). 

Schneider notes that heavily Yankee states 'showed a distinctly liberal trend during the 1964-1972 period', which bears out the idea, discussed in the parent post, that states such as Rhode Island and Connecticut may have a large discrepancy between their #1 and #8 vote shares because there was only one Democratic landslide (1964) after they began turning from Republican into Democratic states. (Although Schneider's comment that McGovern 'did better than his nationwide percentage in all the New England states, including Vermont, New Hampshire, and Maine', does not appear to be correct--McGovern got 36.47% in Vermont and 34.86% in New Hampshire, whereas he got 37.52% nationally. That said, previous Republican landslide winners would have been outstripping their nationwide percentage in Vermont by much more than Nixon did, and the same could be said of North Dakota: Nixon did only 1.50% better in North Dakota than he did nationally, actually a little less than the 1.99% better in Vermont than he did nationally, although archconservative Schmitz got 2.01% in North Dakota and wasn't on the ballot in Vermont.)

-It doesn't seem as though there are any states outside the South where all of the top eight vote shares ever belonged to Democrats, at least in the postbellum period. However, there are six Mountain West states in which Democrats held at least five of the eight highest vote shares as late as at least October 1952. These five states are what Kevin Phillips specifically called the 'interior plateau states' (except less Idaho and plus Montana and Washington). In all of them save Nevada and Washington, Wilson's '16 vote share was one of the top eight as late as October 1952.

--In Arizona, Democrats held five of the top eight vote shares as late as October 1972, and Wilson's '16 vote share was #7 as late as October 1956, and #8 as late as October 1972.

--In Montana, Democrats held six of the top eight vote shares as late as October 1984, and five, as late as October 2004. Wilson's '16 vote share was #8 as late as October 1952, and Bryan's in 1900, #8 as late as October 1984.

--In Nevada, Democrats held five of the top eight vote shares as late as October 1972. Here, Wilson's '16 vote share (53.36%) was never amongst the top eight; these five Democratic vote shares were set by Franklin Roosevelt and William Jennings Bryan. However, two of them belonged to Bryan: his '96 vote share (which is not unusual), and also his '00 vote share. Nevada is the only state outside the South in which Bryan's '00 vote share is still one of the top eight. 

--In New Mexico, Democrats held six of the top eight vote shares as late as October 1952, and five as late as October 1984. Wilson got only a bare majority in 1916, but, in this young state, it was still #8 as late as October 1952 (which, as young as the state was, means he outdid someone who did better than him nationally in the state--in this case, Coolidge). In this state, Harry Truman's '48 vote share was #8 as late as October 1984. In New Mexico, of course, Democrats are back to holding a majority of the top eight vote shares after 2008.

--Washington is a somewhat unusual case, in that after the 1944 election, Democrats have always held a majority of the top eight vote shares in the state. Wilson won Washington with only a plurality in 1916, and his vote share ceased being one of the top eight ever in the state after the 1932 election. Despite its geographic location (and despite that Bryan's '96 vote share stood as one of the top eight earned in the state up until October 2008), Washington's being a majority-top-eight-Democratic state might have more (or as much) to do with its shared heavily Scandinavian heritage with Wisconsin and Minnesota (two states whose runs as all-GOP-top-eight states ended permanently with FDR). Bill Schneider termed Wisconsin, Minnesota, and Washington ('particularly their poorer, rural, Scandinavian areas') as a third, weaker 'core [area] of New Deal Democratic strength' (along with the Deep South and urban Catholic New England). 

Washington also shares with Wisconsin and Minnesota that the very highest vote share earned in the state remains Republican: Theodore Roosevelt's in 1904 (also the top-ever vote share earner in Minnesota; in Wisconsin it is Warren Harding). In the other five Mountain states under discussion, the top-ever vote share is Democratic: Bryan's in '96 in Montana, Utah, and Nevada; FDR's in '32 in New Mexico; FDR's in '36 in Arizona. This may not be saying much, since New Mexico and Arizona achieved statehood after TR's '04 run, and given the idiosyncratically high support Bryan got in the Mountain West in 1896. But in Montana and Nevada, not only is Bryan first, but FDR is second (second and third, in the case of Nevada), ahead of TR (who doesn't even appear in the top eight in either state). In Utah, FDR in '36 was second up until October 1980, and hence ahead of every Republican who ran before him (which would include Harding and TR).

---

Aside from these six states, Wilson's '16 vote share was in the top eight in Wyoming up to October 1952, and in the top eight in Idaho up to October 1936. It was in the top eight in Nebraska up to October 1928, which is not that long, but Nebraska is older than most of the Mountain West states (having first participated in 1868), and was also profoundly ancestrally Republican.

Finally, Wilson's '16 vote share remains in the top eight in Colorado today--the only state outside the South in which this is the case.

-Outside the South and these Mountain West states, Democrats held five of the top eight vote shares in New Hampshire and New Jersey as late as October 1896. I didn't do every timeframe for every state, but those are the only cases I can find of Democrats holding a majority of a state's top eight vote shares between the War of Sections and the 21st century. 

New Hampshire's and New Jersey's situations differed. In New Hampshire, it seemed to be a case of the antebellum Democracy having been very powerful in the state, and the GOP having been somewhat weaker there than elsewhere in New England. Every Democratic vote share in the top eight in the state from December 1872 to October 1896 belonged to an antebellum Democrat (and furthermore to a Democrat whose principal opponent was a Whig). New Jersey, on the other hand, had been a Whig-leaning state during the Second Party System--it had voted Whig in every election from 1836 through 1848. Franklin Pierce did set the all-time high vote share in the state, which stood up to October 1872--but at that time, only four of the top eight vote shares were Democratic. Democrats gained a majority of the top eight vote shares in the state after the war, and as of October 1896, of the five Democratic vote shares in the state's top eight, four were attributable to candidacies from 1860 on.

-In Delaware, Horatio Seymour's 1868 vote share is actually one of the top eight still today. However, we determined that some of Seymour's vote shares in some states are somewhat suspicious. However, as late as October 2008, Grover Cleveland's '88 vote share was in the top eight in Delaware. I haven't necessarily been comprehensive, but I'm pretty sure there are no states outside the South where any Democratic nominee between 1864 and 1892 (inclusive) is still in the top eight, or has been at any point in the recent past. (Democrats also held a majority of the state's top eight vote shares as late as October 1920, which would be highly unusual for a non-Southern state, and particularly for a non-Southern and non-Mountain West state.)

-In Maryland--one of the Southern states where Nixon '72 is the only Republican in the top eight--unlike in most other Southern states, most of the Democratic vote shares above Nixon's are recent (although some are not--FDR's in '32 and '36, Seymour's in 1868). In fact, between Maryland's ancestral Whiggishness during the Second Party System, and its relatively weak allegiance to the Solid Democratic South (a weak allegiance shared, incidentally, with West Virginia), Democrats gained a majority of Maryland's top eight vote shares only with FDR. However, aside from Seymour, Tilden's vote share was in the top eight in Maryland as late as October 1972. 

(In West Virginia, on the other hand, Democrats lost a majority of the top eight vote shares after the 1896 election, and regained it only briefly between 1964 and 1972.)

-Today, a map of the states coloured in by which party has won a majority of its top eight vote shares would look like this. 

-It's particularly interesting to note the top-eight vote shares set by narrow-election or badly losing presidential nominees. 

Out of 1916, 1948, 1960, 1968, and 1976, only one nominee in each of those elections is still responsible for a top-eight vote share in at least one state (Wilson in Colorado, Truman in Missouri, Kennedy in Rhode Island, Humphrey in Rhode Island and Hawaii, and Ford in Alaska). (Alaska and Hawaii are young states, but the longer Ford's and Humphrey's vote shares stand as one of the top eight in them, the more impressive they become.)

In 1916, 1948, and 1960, this was the winning nominee; in 1968 and 1976, this was the losing nominee. Particularly in the cases of 1916 and 1948 (given how long ago they were), it seems there might be a correlation between setting a long-lasting top-eight vote share, and winning in a nationally close election.

However, Charles Evans Hughes set a top-eight vote share in three large states (including the largest and second-largest): New York, Pennsylvania, and New Jersey. They didn't last very long (until October 1928 in New Jersey, October 1924 in Pennsylvania, and only until the next election in New York), but this still seems fairly impressive, given how dominant his party was at the presidential level before and after that election. (At the same time, of course, it bears noting that Wilson set top-eight vote shares in a number of states [outside the South], several of which remained in those states' top eight until October 1952, and at least one of which other than Colorado remained in the state's top eight until October 1972.)

In 1960, Nixon earned the fifth-highest ever vote share in Oklahoma at the time; it stayed in the state's top eight until October 2000. That's fairly impressive.

In 1976, Carter earned the seventh-highest vote share ever won in (then-6-EV-)West Virginia, which remained in that state's top eight until October 2016. (Ford also won the fifth-highest vote share ever won in Utah, which remained in that state's top eight until, I believe, October 2000.) So these elections were fairly balanced, I think it's reasonable to say.

In 1948, on the other hand, I haven't been able to find any state where Dewey's candidacy that year held a top-eight vote share in any state, however briefly. I might have missed one, but in general, he did worse in the interior states than he and Willkie had done in 1944 and 1940, respectively; and he generally underperformed Hughes (let alone some Republican landslide winners) in the East Coast states. Conversely, not only is Truman's vote share that year still in the top eight in Missouri, but Truman's vote share stayed in the top eight in New Mexico down to October 1984; in Oklahoma, down to October 2012; and in West Virginia, down to October 2012.

Of course, in 1968, again, I might have missed something, but it seems that Humphrey set top-eight vote shares in two states that are still standing, whereas Nixon, that year, did so in none, but Nixon won. 

In 2000 and 2004, George W. Bush set top-eight vote shares in a significant number of states (these in 2000 and these in 2004, although they are still in the top eight in these states for 2000 and in these states for 2004). Of course, Al Gore set the fourth-highest vote share in New York in 2000, and it's still in the top eight today. (John Kerry, as discussed, set what was the eighth-highest vote share at the time in 2004, although it was instantly superseded--although I haven't even been able to find anything of the sort for Dewey in 1948. Al Gore and John Kerry both entered into Hawaii's top eight vote shares at the time of their elections--Kerry with less than 55%--but neither is there any longer, whereas Humphrey still is.) 

Taking, again, New York's size into account, it seems that both nominees set an unusual number of top-eight state vote shares in 2000, given the nature of the national election, and certainly that George W. Bush did in 2004. 

In 2012 and 2020, it seems that both nominees have been setting top-eight vote shares in a large number of states for a close election. Romney did so in six states (they are still in the top eight in four states). For the other five nominees in the last three elections, I didn't look for states where they were in the top eight but no longer are; Obama '12 is currently in the top eight in five states (including both New York and California); Trump '16, in six states; Trump '20, in eight states; and Biden '20, in seven states. Hillary Clinton currently is in 'just' two states (one of which is California), but she is already just sixth in both.

Then there are badly-losing nominees who set a top-eight vote share in a state. John Frémont's vote share in 1856 is still the sixth-highest ever in Vermont, and the eighth-highest ever in Massachusetts. He seems very unlikely to be dislodged from his position in the former in the foreseeable future. 

Bryan '00's vote share is still in the top eight in Nevada. (I'm essentially discounting the South for the century after the War of Sections, unless there's something unusual. I suspect the number of states in which Alton Parker still holds one of the top eight vote shares is far out of proportion to the degree to which his candidacy was compelling or captivating.)

Wendell Willkie's vote share in South Dakota was in the top eight until October 1956, and Thomas Dewey's in 1944, until October 1980. Both were also above 55.5%. (Willkie's vote share in North Dakota was in that state's top eight until October 1952--but the Dakotas are both young enough that it seems a vote share should stay correspondingly longer in the top eight before it can be considered impressive. His vote share in North Dakota was 55.06%.)

Dukakis' vote share in Hawaii was in that state's top eight until October 2012. Hawaii is a young state--and Dukakis' vote share was just 54.27%--but that's still 24 years (or that's still five subsequent elections that its place in the top eight survived). He did slightly better in Hawaii than John Kerry did, and substantially better than Jimmy Carter did in 1980, despite that Carter won the national election.

McCain's vote share in Oklahoma is currently the 6th-highest ever achieved in that state. It was 5th as of December 2008. He did better than Bush in '04 (and Bush in '00), and Trump failed to outdo him in the state in either of his elections. 

---

Anomalies

Here and here, I discussed some states I felt were anomalies. 

-Martin van Buren's 1836 vote share remained in several states' top eight for some time after the War of Sections. In Maine, next-door to New Hampshire, it did so until October 1904. However, already by December 1888, the differential between the highest and eighth-highest vote shares in Maine had been whittled down to 11.22%.

Van Buren's '36 vote share hasn't been in any state's top eight--other than New Hampshire (where his is still the all-time highest vote share)--since 1948, when it was pushed out of the top eight in Missouri by Truman. 

New Hampshire was a Democratic state in the 1836-1852 period. But van Buren outdid all of Polk, Cass, and Pierce in the state by a minimum of 18%. (In Maine, he did a little over five points better than Polk, the second-best performing Democrat in the state in the 1836-1852 period.)

Sometimes a nominee is very popular in a region in one election, much more so than any other nominee of his party had been or would be (e.g., Bryan in the Mountain West in 1896). But New Hampshire isn't a region--it's a state. It's hard to see how one conceives of any region that New Hampshire belongs to but Maine doesn't. (And whilst Vermont remained idiosyncratically--relative to the rest of the Yankee Kingdom--Whig-leaning in the 1836-1852 period, for which there appears to have been no really satisfying reason, Maine was clearly Democratic-leaning during this same period--which makes it harder to explain New Hampshire's idiosyncratically towering vote for van Buren in 1836.)

As late as October 1928, the differential between the highest and eighth-highest vote shares in New Hampshire was 18.11%--bigger than in all but five states today. It closed some with Republican landslides between 1928 and 1984 (inclusive), but is still outlierly today. 

-It seems fairly clear that vote shares in the 1864, 1868, and 1872 elections in the Southern states can't be trusted as being all that reflective of a stable electorate. The gap between the highest and eighth-highest vote shares in Kentucky didn't close to under 20% until after the 1928 election. In Missouri, it was never quite as high, but, for example, as of October 1928, over 2/3 of the 14.26% differential between the highest and eighth-highest vote shares in Missouri was accounted for by the difference between the highest (Lincoln's in '64) and the second-highest (van Buren's in 1836). 

In Tennessee, the differential between the highest and eighth-highest vote shares remained over 20% as of October 1972.

There are two states I mentioned as not showing up as particularly outlierly today here: Maryland and West Virginia. In Maryland, the differential between the highest and eighth-highest vote shares was, perhaps, not enormous for the time (one would expect this differential to be larger, the farther back one goes), but, as of October 1932, over 3/4 of the 13.05% differential between the highest and eighth-highest vote shares in the state was accounted for by the difference between the highest (Seymour's) and the second-highest (Hoover's). In West Virginia (which was, of course, a relatively young state during the first half of the 20th century), as late as October 1932, 63.5% of the 14.82% differential between the highest and eighth-highest vote shares was accounted for by the difference between the highest (Lincoln's in '64) and second-highest (Grant's in '68). 

As for the idiosyncrasy of Horatio Seymour in general, it seems, again, plausible that Tilden was particularly exciting in the South. Tilden set top-eight vote shares, not just in the South, but also in New Jersey, New York, and Connecticut (admittedly, so did Seymour in New Jersey and New York, albeit lower and shorter-lived ones). And, again, Tilden was the only Democrat between 1852 and 1932 to win a national majority, which it is sometimes easy to forget. According to Justin Gabriel (who refers to They Also Ran's Wikipedia page), 'Tilden was nominated with more enthusiasm by the Democrats than any candidate since Andrew Jackson.'

But, for example, in Delaware, no subsequent Gilded Age Democrat outdid Seymour, and as of October 1920, about 46%--nearly half--of the differential between the highest and eighth-highest vote shares in Delaware was accounted for by the differential between the highest (Seymour's) and the second-highest (Cleveland's in '84). 

My guess would be that there was some mechanism whereby the electorate that supported Seymour in 1868, in those Southern states where his vote share stood for particularly long as the highest or one of the highest, was depressed after that election. (Much as, as was described here, the electorate that supported Grant--and, presumably, Lincoln--was depressed, or offset, between 1868 and 1872 in Missouri.)

1864 also seems like an outlier for Kansas. In October 1952, the differential between the highest and eighth-highest vote shares in Kansas was 17.45%; in Nebraska, it was 9.30%. The difference between the second- and ninth-highest vote shares in Kansas, as of that time, would have been no more than 11.62%.

Of course, a 'stable' electorate doesn't necessarily mean 'representative'. During the war and Reconstruction, Southern states varied wildly in terms of their levels of support for the two parties' nominees, both between elections and amongst one another. This had mostly concluded by 1876, and definitively, by 1880. However, after the Mississippi Plan constitutions took effect, there was another dramatic change--but this time, it was a one-time change leading to a new, stable electorate that appears to have remained in place for several decades. Because the electorate was stable, the vote in the various Southern states was more or less what one would expect it to be from election to election and from state to state, within this period. But, in the grand scheme of things, it made, certainly, the entire pre-Fort Sumter Confederacy an outlier in terms of (for example) the lowest of the top eight vote shares being greater than 9/13.  

Indeed, it's probably fair to say that the top vote shares have been warped in every state where either Alton Parker or John Davis still holds one of the top eight vote shares, even if in some cases (e.g., Virginia, North Carolina) the lowest of the top eight vote shares is not all that high relative to other states. (The 'likely blue' states are ones where Parker still holds one of the top eight vote shares; the 'leans blue' states are ones where Davis does. In South Carolina, both--along with James Cox--do.) Of course, some losing nominees--even badly losing ones--are particularly compelling in some part of the country (Smith in urban Catholic New England, Goldwater in the South or, more specifically, the Old Confederacy, etc.). But there seems to be little evidence that this would have been the case of Davis or Parker (or Cox)--at least, not to that extreme a degree. 

A good comparison would be with the 1952 and 1956 elections, when Stevenson did carry a number of Southern states--but his best vote share in any state in '52 was 69.66%, in Georgia; and in '56, 66.48%, in Georgia. That's high, but in 1904, Parker won five states with a higher vote share than Stevenson carried Georgia with either time, and carried his best state by better than 21:1; in 1924, Davis won five states with a higher vote share than Stevenson carried Georgia with either time, and won his best state by better than 21:1; and even in 1920, Cox carried three states with a higher vote share than Stevenson carried Georgia with either time, and carried his best state by better than 24:1. Another comparison could be with the close elections of the Gilded Age; one would certainly have expected Tilden, Hancock, and Cleveland to have been more compelling to the South than Parker or Davis. (Or to antebellum elections, where one would have expected some of the nominees--depending on whether the state was Whig- or Democratic-leaning--to have been more compelling than Parker or Davis; indeed, in Texas, Pierce is still in the top eight vote shares, although he is still behind Davis.)

Or, again, George Wallace seems to have been a far more compelling figure in US political history than Strom Thurmond; and yet, Thurmond carried a higher percentage of the vote--considerably higher--in three states, than Wallace did in any state. 

Even at the county level, in 2020, Trump got 96.18% of the vote in Roberts County, Texas, which was his best county according to Leip. (Biden didn't get over 90% in any county; his best county was Prince George's, MD, where he got 89.26%.) Trump's vote share in Roberts County was actually lower than Davis' in the entire state of South Carolina (96.56%), and only slightly higher than Parker's (95.36%) or Cox's (96.05%) in that state. Roberts County cast just 550 votes in 2020, but Trump couldn't win it better than Davis won the state of South Carolina. (Each of Trump's seven best counties--Roberts, TX, Borden, TX, King, TX, Garfield, MT, Glasscock, TX, Wallace, KS, and Grant, NE--cast fewer than 1,000 votes in 2020. Trump's best county that cast over 1,000 votes was Armstrong, TX, which cast 1,112 votes, and where he got 93.08%--worse than any of Parker, Cox, or Davis in South Carolina. And apparently, Trump's performance in Roberts County was the best of any major party nominee in any county in the last six elections--the best Democratic county performance was Obama's in Shannon, SD in 2012, where he got 93.35%.)

More generally, as mentioned previously, the highest vote share any nominee has received in a state  outside the South is Bryan's 84.95% in Colorado in 1896. The highest vote share received by a Republican in any state outside the South (which is to say, excepting Goldwater in Mississippi) is McKinley's 80.08% in Vermont. There are a number of vote shares close to these in other non-Southern states, won by nominees for which these vote shares in those states make sense (Bryan's 82.70% in Utah, 81.21% in Nevada, 79.93% in Montana, and 78.10% in Idaho in 1896; Johnson's 78.76% in Hawaii and 80.87% in Rhode Island in 1964; Harding's 77.79% in North Dakota in 1920; and Grant in both '68 and '72, Coolidge, TR, and Frémont in Vermont--these are all the cases [excepting Kansas in 1864] outside the South, or, for that matter, inside the South either since 1972 or before 1888, in which a nominee carried a state with over 77.7%, winning it by better than 3.5:1). 

(FDR's best non-Southern vote shares in 1936 were 72.81% in Nevada, 69.85% in Arizona, 69.34% in Utah, and 69.28% in Montana; in 1932, when he did atypically well in the Plains West, they were 69.59% in North Dakota, followed by 69.41% in Nevada and 67.03% in Arizona. By comparison with some other landslide winners who won no states by better than 3.5:1, Hoover got 72.02%, 70.36%, and 68.63% in his three best states [Kansas, Michigan, and Maine--Vermont was fifth] in 1928; Eisenhower got 72.16%, 70.87%, and 66.11% in his three best states [Vermont, Maine, and New Hampshire] in 1956; and Reagan got 74.50%, 72.36%, and 70.55% in his three best states [Utah, Idaho, and Nebraska] in 1984.) 

(Nixon should probably get 'credit' for his 78.20% win in Mississippi in 1972; it's not easy to tell when the 'Mississippi Plan era' ended in any given state, but, unlike Johnson in 1964, McGovern carried three counties in Mississippi, all of which have gone on to vote Democratic in every subsequent election. He likewise carried six in Alabama, two in North Carolina, and one in Virginia, all of which have gone on to vote Democratic in every subsequent election. But he carried no counties in Georgia, South Carolina, or Arkansas; and in Louisiana, he carried one parish, but it has voted Republican in the last six elections. So in Mississippi and Alabama, at least, the faint outlines of the ancient Cretaceous shoreline through the South are visible in 1972, which can be used as a rough index of when the 'Mississippi Plan era' ended in a given state. [And, while McGovern's three counties in Mississippi are many fewer than for any subsequent Democrat, he was especially unpopular amongst many traditionally Democratic constituencies in a way that none of the next five or six Democratic nominees was; Mondale held more ground in Mississippi than McGovern did, but the same was true of rural areas of Tennessee, Kentucky, West Virginia, and Pennsylvania that have turned bright red in the 21st century, and of some large, urban counties that have been blue ever since their Nixon '72 vote, such as Cuyahoga, OH, Allegheny, PA, Lake, IN, and Jackson, MO.])

Removing the Mississippi Plan-era South, the highest vote shares received by the Democracy and the GOP in a state are comparable, and occurred in comparable contexts (indeed, in the same election). The other highest vote shares received by the respective parties are also comparable and were achieved in elections, and by nominees, who made sense (even if not all of them were winning national landslides, or even winning at all). Leave the Mississippi Plan-era South in, and this is not the case.

Hence the attempted 'normalisation' of the South here (although, as discussed there, there's not really any good way--or at least, not one I know of--to really 'normalise' the South in a way that makes sense--e.g., that doesn't simply dismiss FDR's [or Wilson's] vote shares in the South altogether). It's just useful to keep in mind.

In fact, we could even try designating a 'Most Democratic State' outside the former Confederacy (and/or pre-Fort Sumter Confederacy) from 1884-1944 (although this, again, shouldn't be taken as the 'actual' Most Democratic State in those elections, since, under normal conditions, the Most Democratic State might well have been in the former Confederacy/pre-Fort Sumter Confederacy--aside from which, it's likely meaningful that the Democratic vote share in South Carolina nearly consistently managed to outpace that in other Deep South states):

1884: Arkansas [57.83%]/Delaware [56.55%]/Indiana [49.46%]
1888: Delaware [55.15%]/New Jersey [49.87%]
1892: Arkansas [59.30%]/Maryland [53.39%]/New Jersey [50.67%]
1896: Colorado [84.95%]
1900: Arkansas [63.46%]/Nevada [62.25%]
1904: Virginia [61.84%]/Kentucky [49.82%]/New York [42.28%]
1908: Virginia [60.52%]/Kentucky [49.74%]/Nebraska [49.14%]
1912: Virginia [65.95%]/Maryland [48.57%]/Nebraska [43.69%]
1916: Virginia [66.99%]/Colorado [60.74%]
1920: Virginia [61.32%]/Kentucky [49.69%]/Arizona [44.39%]
1924: Virginia [62.48%]/Oklahoma [48.41%]/New Mexico [43.02%]
1928: Arkansas [60.29%]/Massachusetts [50.24%]
1932: Arkansas [85.96%]/Oklahoma [73.30%]/North Dakota [69.59%]
1936: Arkansas [81.80%]/Nevada [72.81%]
1940: Arkansas [79.02%]/Arizona [63.49%]
1944: Arkansas [69.95%]/Utah [60.44%]

(I subsequently added a third category--for best state outside the former slave states [including Oklahoma] altogether, when applicable. In 1892, Cleveland won a majority in both New Jersey and Connecticut, the first time since 1880 [when Hancock won Nevada with a majority] that the Democrat had won a majority in a free-soil or postbellum state.)

Here, only FDR's '32 vote share in Arkansas is higher than what Bryan got in Colorado in 1896--and only slightly so. There is a certain interesting regularity here, although I don't know what that is down to--it is possible of course that voting restrictions differed amongst the post-Fort Sumter Confederate states. In Arkansas, there are a suspicious number of counties that voted Republican several times in a row in the late 1800s, that then switched and thereafter voted continuously Democratic for several decades (Desha, Lincoln [except in 1920], Little River, Phillips, Pulaski). In Tennessee, there is only one such county (Fayette). (In Virginia, there are also a number of such counties--Buckingham, Cumberland, Elizabeth City County, Henrico [except in 1928], James City County [except in 1928], King William, Lancaster [except in 1928], Louisa [except in 1928], Lunenburg, Mecklenburg, Middlesex, New Kent [except in 1928], Northampton, Powhatan, Prince Edward, Prince George, Richmond County, Sussex, Westmoreland [except in 1928], Williamsburg City. In Kentucky and Maryland, I would say there are none, and in Missouri, the closest you get are some counties that voted Republican in 1864, 1868, and 1872, before beginning a long Democratic streak.)

However, there were six counties in Arkansas that cast less than 0.5% as much vote as Los Angeles County (the country's biggest county) in 2020, that Biden carried; there was only one such county in Tennessee (Haywood). So perhaps we shouldn't expect there to be as many, or any, suspicious counties in Tennessee. (The six Biden counties in Arkansas that cast less than 0.5% as much vote as LA County included Desha and Phillips. These were the states that included any such counties in 2020, with Mississippi leading the way with 27, followed by Georgia with 14, Colorado with 13, and Alabama with 10 [Virginia had 24 total counties and independent cities that cast less than 0.5% as much vote as LA County that voted for Biden, but only seven counties proper]. These were the states with [to take an arbitrary number] at least five such counties in 2020. This was, obviously, meant as something of a blunt way of measuring the prevalence of this phenomenon--and similar phenomena in the Southwest and northern Plains and Mountain West--which it does somewhat but imperfectly, especially as there are increasing numbers of small 'ski resort and mountain nirvanas' and 'lifestyle communities' being carried by Democrats.

Maryland's apparent belonging to the South here, I think, is misleading. Biden's one < 0.5%-of-LA County county in Maryland was Kent; I'm not sure what the explanation for his flipping Kent County, MD was, and the most I've found is a reference to 'many [of the larger cities in otherwise Republican-dominated areas being] home to college campuses'. [Chestertown was one of these 'larger cities' listed, but it wasn't specified whether it was one of those that was home to a college campus.] It wouldn't have been, however, in 2012, when Obama carried Somerset County. Of course, as late as 2012, there was also a much broader range of < 0.5%-of-LA County blue counties; these were the states where he carried at least one such county in 2012. There were 22 in Iowa alone, including Howard. Kentucky is blue on this map only because of Elliott County. But, Pennsylvania--which had 13 < 0.5%-of-LA County counties in 2012, one of which [Elk] had voted for Obama in '08, and another of which [Greene] had voted for Gore in '00--still forms something of a barrier between Maryland and the non-'border North'.)

-Of course, to see how outlierly a state is at a given point in time, one should compare it with what appears to be the norm at that time. As mentioned above, one would generally expect the differential between the highest and eighth-highest vote share--and also the differential between the highest and second-highest vote share--to be greater in any given state, the farther back one goes (and the fewer elections it has taken part in). I compared New Hampshire to Maine (its neighbour, and a politically closely-aligned state at least during the 1836-1852 period); and Kansas to Nebraska. I didn't compare some of the border states whose top vote share was (or was until recently) set in the 1864-72 period with anything else. 

But, for example, if we look at Connecticut as of October 1920, the differential between the highest and eighth-highest vote shares was 10.83%, and the differential between the highest and second-highest vote shares accounted for about 35.2% of this. The corresponding figures for some other states around the same time:

Indiana, October 1920: 4.88%, 2.25%
Iowa, October 1952: 12.87%, 52.76%
Massachusetts, October 1920: 14.01%, 17.56%
Michigan, October 1920: 13.29%, 51.54%
Michigan, October 1964: 18.14%, 14.39%
Minnesota, October 1920: 14.92%, 70.04%
Minnesota, October 1932: 14.87%, 22.80%
New Jersey, October 1928: 13.25%, 41.36%
New Jersey, October 1964: 10.85%, 27.37%
Ohio, October 1936: 11.65%, 44.12%
Ohio, October 1964: 8.52%, 44.37%
Pennsylvania, October 1920: 15.03%, 39.45%
Pennsylvania, October 1964: 9.16%, 24.45%
Rhode Island, October 1920: 11.17%, 32.32%

I mentioned the overall gap between the highest and eighth-highest vote shares being over 20% in Kentucky until the 1928 election, and in Tennessee until the 1972 election. Out of the states above--and I made sure to include some of the quasi-outlierly Harding/TR-record-vote share states--none of them had a differential of > 20% as of the stated points in time. (And only Michigan as of October 1964 had one that was greater than the 18.11% differential in New Hampshire as of October 1928.)

I mentioned the gap between the first- and second-highest vote shares being responsible for over 2/3 of the overall differential between the first- and eighth-highest in Missouri as of October 1928; over 3/4 of the overall differential in Maryland as of October 1932; 63.5% of the overall differential in West Virginia as of October 1932; and about 46% of the overall differential in Delaware as of October 1920. Of the cases listed above (which are, of course, not comprehensive, the gap between the first- and second-highest vote shares is responsible for over half of the overall gap only in Iowa as of October 1956, Michigan as of October 1920, and Minnesota as of October 1920. (It was higher than the proportion accounted for in Delaware in no additional states.) In Michigan and Minnesota, the proportion accounted for by the gap between the highest and second-highest vote shares diminished to under a quarter relatively rapidly (by October 1964 in Michigan). In Missouri, this proportion was still 46.07% as of October 1948; in Maryland, 40.05% as of October 1952; and in West Virginia, 59.35% as of October 1964.

(That said, this proportion was 65.35% in Wisconsin, a Harding-record state near the end of the alphabet, as of October 1964.)

But again, the top vote share in Missouri and West Virginia had been set by Lincoln, who belonged to the opposition party to that for which these states would vote in every election from 1872-1900 and 1872-1892, respectively; and in Maryland, by Horatio Seymour (whom, for some reason, every subsequent Gilded Age Democrat would substantially underperform--and who lost decisively nationally). In Wisconsin, it was set by a landslide-winning Republican in what was clearly at the time a traditionally Republican state, and who was doing particularly well across the Upper Midwest, the region in which Wisconsin lies (which could, again, also be said of Iowa, Michigan, and Minnesota as well). 

---

Relatively non-partisan/partisan states

Of course, the original reason I chose 'eight' was because that was the minimum number of times that every state has now been carried with at least 55.5% of the vote. It so happens that, as of October 2000, eight was also the minimum number of times that every state except Alaska and Hawaii (the two youngest by far) had been carried with at least 55.5% of the vote. (But this hadn't been the case before the 1996 election.)

Some states got all their top eight vote shares above 55.5% faster than others. Some states that did so relatively quickly were Arizona, California, Idaho, Iowa, Maine, Massachusetts, Minnesota, Montana, Nebraska, Nevada, North Dakota, Oklahoma, Rhode Island, South Dakota, Texas, Utah, Vermont, Washington, and Wyoming (and, for that matter, Alaska and Hawaii). (Texas got there at a time when its three top vote shares were all set by antebellum nominees; for the most part I am discounting the South or at least the pre-Fort Sumter Confederacy.) 

What were the states that were particularly slow to get to having all their top eight vote shares being above 55.5%?

-Connecticut, an original state, didn't get there until Dec. 1956.
-Delaware: sometime between 1972 and 1988
-Illinois: sometime between 1932 and 1956
-Indiana: sometime after 1964 (as of October 1964, two of its top eight vote shares were still below 55.5%)
-Kentucky: sometime between 1932 and 1936
-Maryland: December 1964
-Missouri: December 1928
-New Hampshire: December 1920 (quite at odds with van Buren's 75.01% there)
-New Jersey: December 1952
-New York: December 1996
-Ohio: sometime between 1936 and 1956
-Oregon: sometime between 1932 and 1952
-Pennsylvania: sometime between 1920 and 1928
-Tennessee: sometime between 1932 and 1944
-West Virginia: sometime between 1964 and 1976
-Wisconsin: December 1932

It seems like North Carolina might also have taken relatively long to get there. 

This is an impressionistic list, but a map of the above states would look like this. What do we see? The states are overwhelmingly in the upper right quadrant of the country. Looking more closely, however, most of New England, and most of the Upper Midwest, are left blank. Most of the former Confederacy is left blank, but the border states (including Tennessee--and, possibly, North Carolina) are all states (well, except Oklahoma) that took a while to get to this place (even with whatever restrictive voter laws they may have passed, in the case of those states that formerly belonged to the Confederacy). And then every Route 30 state (and every 'mean centre' state) is also on this map. 

Bigness seems correlated with tending not to vote in extreme ways. New York, Pennsylvania, Illinois, and Ohio are all on this map--these were the four 'leading states' before the rise of the Sun Belt. But there also seems to be a general region of low-partisanship running through the northern and southern border states. This is also where third parties tend to do most poorly. 

New York went the longest without all eight of its top vote shares being above 55.5%, by a fair amount--the runner-up appears likely to have been Delaware, followed by Indiana or West Virginia.

Many of the states that used to be relatively low-partisan have become very high-partisan recently, such as Tennessee, Kentucky, West Virginia, Indiana, Maryland, and New York. On the other hand, Ohio's top eight haven't changed since 1984; New Jersey's, not since 1984; Missouri's, not since 1984; North Carolina's, not since 1984; and Oregon's, not since 1964. It appears that, outside the pre-Fort Sumter Confederacy and Vermont, Oregon, Pennsylvania, and Wisconsin are the states whose top eight vote shares have remained unchanged the longest (since 1964--Vermont's has remained unchanged since 1924).

The states I mentioned, furthermore, are states where one might have expected recent additions to the top eight vote shares. Well, perhaps not Ohio (yet), but yes in the other cases. They are all states that have become safe for one party or the other in the last six elections (albeit only extremely recently in Ohio's case), and in which the lowest of the top eight vote shares is below 60%. New Jersey is particularly striking, because it is the only one of the seven states that I proposed here as a specific definition of 'urban coastal America' in which no recent Democrat has broken into the top seven. (The closest was Obama's 58.38% in 2012, which still fell over 1% short of McKinley's 1896 vote share.) In Oregon, an increasingly Democratic state, the highest vote share out of the last six elections was Obama's 56.75% in 2008 (here, unlike in Washington, Biden failed to outdo him, getting 56.45%). Obama's vote share was a little over 1% short of FDR's 1932 vote share in the state; Biden's was over 1.5% short thereof. 

Another thing about the Route 30 states is that their top eight vote shares tend to be disproportionately earned by national landslide winners. (By 'landslide', in this paragraph, I mean, someone who is getting at least 55.5% of the national vote.) In Illinois, six of the top eight vote shares belong to national landslide winners (three of whom were doing worse in the state than they were doing nationally); the other two belong to Calvin Coolidge and Obama in '08. In New Jersey, six of the top eight vote shares belong to national landslide winners (all of whom were doing better in the state than nationally); the other two belong to Calvin Coolidge and McKinley '96. In Ohio, seven of the top eight vote shares belong to national landslide winners (two of whom were doing worse than they were doing nationally); the eighth belongs to Calvin Coolidge. In Pennsylvania, five of the top eight vote shares belong to national landslide winners (all of whom were doing better than nationally); the other three belong to Coolidge, McKinley '96, and McKinley '00. (At this point, I guess it seems awkward to continue to maintain that Coolidge was not a landslide winner, but after all, his vote share was lower than FDR's in '40, and you don't see FDR in '40 in any of these.) 

In Indiana, the most partisan (at least recently) of the Route 30 states, only four of the top eight vote shares belong to landslide winners (none of them--unlike in the other states--a Democrat; all of them doing better than they were doing nationally). A fifth belongs to Eisenhower in '52, who was a landslide winner according to the 'official' definition that appears to have been settled on in the 1980s by serious political scientists, before the term lost all meaning (and who, again, got a higher vote share than Coolidge). The rest belong to Bush Sr in '88, George W. Bush in '04, and Trump in '20. Indiana is the only Route 30 state in which a loser (Trump in '20) holds one of the top eight vote shares. 

This was also largely true of New York between 1972 and 1996 (when all of its top eight vote shares were above 55%, if not above 55.5%). In that period, five of its eight top vote shares belonged to landslide winners (two of whom were doing worse than nationally); a sixth belonged to Eisenhower '52; a seventh belonged to Coolidge '24; and an eighth belonged to McKinley '96. Before 1972, one belonged to van Buren '36; after 1996, one belonged to Clinton '96, who wasn't even winning a national majority. 

Of course, it doesn't take much for a national landslide winner to set a high vote share in a state--he needn't even necessarily be doing better in the state than he is doing nationally, and could even be doing a little worse. 

On the other hand, in the states that relatively rapidly rose to having all eight top vote shares above 55.5%, there were many cases of national losers or narrow-election nominees doing much better than nationally. Franklin Roosevelt did much better in Arizona than he did nationally all four times, as did Woodrow Wilson in 1916. In California, a series of landslide winners (plus FDR in '40 and Coolidge in '24) did much better than they did nationally. In Iowa, Hayes and Garfield were in the top eight when the state first reached all-eight-above-55.5%-status. In Maine, Frémont, van Buren '36, Harrison '88, and Hayes '76 were; in Massachusetts, Frémont, Garfield, and Hayes; in Michigan, Taft (not a narrow-election nominee, but certainly no landslider), Frémont, and van Buren '36 (and Lincoln '60, who was getting less than 40% of the national vote, although he did win the North); in Minnesota, Garfield (along with Taft and Lincoln '60); in Montana, Bryan in both '96 and '00, Wilson in '16, and FDR in '40; in Nevada, Bryan in both '96 and '00, Weaver, Harrison '88, and Blaine; and so forth. 

Wisconsin was a little different; it reached all-eight-above-55.5%-status in December 1932. That doesn't seem like it took that long. At that time, three of its eight top vote shares belonged to landslide winners, with the other eight belonging to McKinley '96, McKinley '00, Lincoln '60, Grant '68, and Lincoln '64. Well, perhaps it wasn't that different. No Republican nominees between Hayes and Harrison (inclusive) were included, and neither was Frémont (who had to drop off the list before Wisconsin reached all-eight-above-55.5%-status), or any other losers. But that is not that different to Arizona or California. I suppose the difference is that, more specifically, it differs from its fellow then-traditionally Republican states in the Upper Midwest (Iowa, Minnesota, Michigan).