Most polarizing election in the past 70 years

[SingingAnalyst]

Very interesting question! I would rank them as (1) 2016; (2) 1968; (3) 2004-- though I also think a good case could be made for reversing the order of the top two.

I looked mainly at data visible on this site:

(1) County percentages:

In 1968, Nixon only received 80.04% or more in 5 counties: his top county was Hooker NE 87.94% and his 5th best Sioux IA 80.04%-- and Nixon's top 5 county percentages came from 5 different states (4 in the Midwest Plains and 1 in Appalachia).

In 1968, Humphrey received 75.62% or more in 5 counties: his top county was Duval TX 88.74% and his 5th best Suffolk MA 75.62%. 3 of Humphrey's 5 best counties were in TX.

In 1968, Wallace received over 90% in two counties: Geneva AL and George MS. Neither Nixon nor Humphrey reached 90% in any county.

Looking beyond the county percentages, however, Black precincts went overwhelmingly for Humphrey, though few if any matched county delineations.

In 2016, Trump received 91.62% or more in 5 counties (4 in TX) and Clinton received 86.80% or more in 5 counties (in 5 different "states"). Johnson's highest percentage was 13.89% in Los Alamos NM and Stein's top was Sioux ND 10.39%, so clearly Johnson's and Stein's percentages do not show a clear geographic pattern as did Wallace's in 1968.

Based on county percentages, I call 2016 and 1968 a tie.

2. County totals:

In 1968, Nixon's top 5 counties gave him 3,178,000 votes (10% of Nixon's total) and Humphrey's top 5 gave him 4,073,000 votes (13% of Humphrey's total)-- indicating a pro-Humphrey bias among the largest counties. Wallace's top 5 counties only gave him 636,000 (6.4% of Wallace's total)-- an indication his strength lied elsewhere.

In 2016, Clinton's top 5 counties gave her 6,237,000 votes, or nearly 10% of her total-- while Trump's 5 largest counties gave him just 3,048,000-- or less than 5% of his total. This shows a very strong pro-Clinton bias among the largest counties.

Based on county totals, I call 2016 more divisive.

Overall, I would rank 2016 slightly more divisive than 1968. Much of 1968's division lied outside the partisan political sphere: 1968 anti-Humphrey protesters at the DNC; the belief of many on the left that no candidate was really worth voting for (a sentiment that, while present in 2016, was much weaker).

2004 I would rank 3rd. Unlike 1968 and 2016, it is harder to find subgroups of the population of which it can be said, "almost every [X] voted for ..."

[SingingAnalyst]

One factor that I think takes the polarization of the 1968 election (arguably the most polarized of the 3) down a notch, is that it was won by the candidate who was perceived at the time to be the most centrist: Richard Nixon.

Unlike Humphrey and Wallace, it is hard to find a subgroup of the population in which nearly everyone voted Nixon (86% of Republicans voted for him). It is easy to find places where almost no one voted Nixon: Detroit's 22nd District (96.2% Humphrey); Homles County, MS (52% Humphrey, 41% Wallace, 7% Nixon); and Geneva County, AL (4.4% Humphrey, 3.3% Nixon, 91.7% Wallace). But other than his lack of appeal to Blacks and Jewish voters, Nixon's support was fairly broad (by 1960s standards).

Trump, on the other hand, is viewed as a unifier by almost no one, including probably most of his supporters.