2020 Presidential Election Exit Polls

The CNN exit polls have data broken down by 24 states in addition to a national total for the 2020 Presidential election. 

Since third-party candidates didn't perform strongly in 2020, these results can be summarized well with the percentage of the vote for a single candidate, and in the results below I will use the percentage of a demographic voting for Biden (recognizing that a 50% vote for Biden means a Biden majority of the two party vote in that jurisdiction, because some votes were cast for third parties).

In broad brush, a state's results can largely be explained by the percentage of the white vote going to Biden and by the racial breakdown of the voters in a state.  

According to the CNN exit poll, Biden won every state polled where Biden got 40% or more of the white vote except Iowa where Biden got 43% of the white vote, and voters were 92% white, 2% black, 4% Latino, 1% Asian and 2% other race, while Trump won every state where Biden got less than 40% of the white vote. Biden got only 76% of the small black vote in Iowa, but got 67% of the Latino vote in Iowa.

In the states that turned out to be close in the 2020 Presidential election, Biden's percentage of the white vote was as follows:

Arizona 46% (Biden) 74% white voters.

Wisconsin 46% (Biden) 86% white voters.

Michigan 44% (Biden) 81% white voters.

Nevada 43% (Biden) 65% white voters.

Pennsylvania 42% (Biden) 81% white voters.

Florida 37% (Trump) 62% white voters.

North Carolina 33% (Trump) 65% white voters.

Texas 33% (Trump) 60% white voters.

Georgia 30% (Biden) 61% white voters.

States with that range of percentages of the white vote going for Biden that were not close were:

Virginia 45% (Biden) 67% white voters.

Iowa 43% (Trump) 92% white voters.

Ohio 39% (Trump) 84% white voters.

Montana 39% (Trump) 88% white voters.

Kentucky 33% (Trump) 88% whit voters.

There was not a single state where data is available where Black, Latino or Asian support for Biden was less than 53% and the spread of percentage support for Biden among white voters by state was about twice as great as the spread of percentage support for Biden among the other three categories of non-white voters upon which there is meaningful data.

The National Results By Partisanship

Nationally 54% of independents (26% of those surveyed) favored Biden in the CNN exit poll, 94% of Democrats (37% of those surveyed) favored Biden, and 94% of Republicans (36% of those surveyed) favored Trump. 

In the Associated Press Votecast Exit Polling 52% of independents (5% of  those surveyed) favored Biden as did 95% of those who were Democrats or lean Democrat (47% of those surveyed), while 91% of those who were Republicans or lean Republican (48% of  those surveyed) favored Trump.

Taken together, the key factor in this year's election were independents who lean Republican (about 12% of voters) but voted for Biden in large numbers.

2016 Vote

CNN's exit poll reported that 95% of people who voted for Hillary Clinton in 2016 (40% of the total) voted for Biden while 4% voted for Trump, that 92% of people who voted for Trump in 2016 (43% of the total) voted for Trump while 7% voted for Biden, that 60% of people who voted for someone else in 2016  (5% of the total) voted for Biden while 25% voted for Trump (with only 15% continuing to vote for a third-party candidate), and that 58% of people who didn't vote in 2016 (11% of the total) voted for Biden while 39% voted for Trump. 

Votecast also reported that 96% of people who voted for Hillary Clinton in 2016 (38% of the total) voted for Biden while 3% voted for Trump, that 93% of people who voted for Trump in 2016 (41% of the total) voted for Trump while 6% voted for Biden, that 57% of people who voted for someone else in 2016  (8% of the total) voted for Biden while 28% voted for Trump (with only 15% continuing to vote for a third-party candidate), and that 58% of people who didn't vote in 2016 (15% of the total) voted for Biden while 41% voted for Trump. 

Of course, those numbers also show some notable inaccuracy in the Votecast data, because the percentage of voters who voted for somebody else in 2016 was less than half of the 8% who reported having done so, and the number of voters who voted for Clinton in 2016 was greater than the number of voters who vote for Trump in 2016 by about three percentage points (the reverse of what voters reported this time). It could be that many Sanders voters who voted for Clinton in 2016 inaccurately reported voting for "somebody else" in 2016, perhaps because they didn't understand the question correctly.

The inaccuracy in the relative Clinton-Trump votes cast in 2016 is similar in the CNN poll, although the percentage who said that they voted for a third-party candidate in 2016 at 5% is closer to the real number.

But, the numbers on prior voting in 2016 still tend to support the narrative that independent voters who lean Republican and in many cases voted for Trump in 2016, who voted for Biden in 2020 were decisive in changing the outcome in 2020.

Nationally, broken down by race, the percentage support for Biden was as follows:

Black 87%

Latino 65%

Asian 61%

Other 58%

White 41%

The Associated Press Votecast exit poll results (nationally) were: Black 90%, Asian 70%, Latino 63%, Other 58%, Native American 46%, White 43% and Native Hawaiian and Pacific Islander 35%.

Put another way the racial breakdown of Biden voters (using the Votecast numbers) is:

White 64%

Black 18%

Latino 11%

Asian 3%

Other 4% (including Native American and Pacific Islander)

The racial breakdown of Trump voters (using Votecast numbers) is:

White 87%

Black 2%

Latino 7%

Asian 1%

Other 4% (including Native American and Pacific Islander)

Nationally, by religion the break down (for statistically significant national subsamples) was:

Religiously affiliated non-Jewish, non-Christians 69%

No religion 65%

Catholic 52%

Protestant or other Christian 39%

White Catholic support for Biden (who is himself a white Catholic) was 44%, while Latino Catholic support for Biden was 65%.

The breakdown of the religious non-Jewish, non-Christian vote is tricky to make sense of since it is a heterogeneous category including, for example, Muslims, Buddhists, Hindus, neo-pagans, and religious Unitarian/Universalists. Muslims, in particular, are probably more right leaning than others in this category generally speaking, but Trump's strong anti-Muslim rhetoric may have alienated even Muslims who favor conservative policies on most economic and social issues. Overall, religious non-Christians who were not Jews, and Jews (see below), supported Biden to a similar degree, and non-religious people also supported Biden to a fairly similar degree. 

There were no statistically significant breakdowns available for Jewish voters in the CNN exit polling.

Other exit polls did look at the Jewish vote, which no one disputes favored Biden, although the magnitude of this support ranged from 61% in a GOP poll to 68% in an Associated Press sponsored poll to 77% in a liberal group's poll.

[A] poll commissioned by the Republican Jewish Coalition found that 30.5 percent of Jewish voters voted for GOP incumbent Donald Trump nationally compared to 60.6% for Democratic challenger Joe Biden. . . .

Meanwhile, a poll commissioned by the liberal group J Street found that 77% of Jewish Americans voted for Biden and only 21% for Trump. . . .

[T]he major media consortium that traditionally cooperates in exit polls — the National Election Pool, which includes The New York Times, The Washington Post and CNN, among other outlets — did not post Jewish results this year.

In 2016, the National Election Pool found that 71% of Jewish voters chose Hillary Clinton and 24% opted for Trump. But there were no Jewish results posted this year, The Washington Post said, because there were “not enough respondents to break down details.”

The Associated Press, which once belonged to the consortium, left in 2018 and set up Votecast, which this year did include a Jewish breakdown. It found that 30% of Jewish Americans were voting for Trump and 68% for Biden. That analysis included about 3,300 Jewish voters (3% of the total 110,000 people surveyed).

Votecast includes major shifts in methodology from the National Election Pool exit polling system, including online panels and, in some cases, compensation for participating — departures substantive enough to make it impossible to compare with past National Election Pool polling. . . .

Both the RJC and J Street polls were completed during the same time period, had similar margins of error (3.5-4%) and included hundreds of Jewish voters — 600 for the RJC and 800 for J Street. But the RJC survey included more Orthodox Jews, who tend to vote Republican, and fewer Reform and unaffiliated Jews, who tend to be solidly Democratic, than the J Street poll. That could explain why the RJC poll yielded a result that suggests more Trump voters.

The Associated Press Votecast exit poll results (nationally) were: None 72%, Jewish 68%, Muslim 64%, Something Else 62%, Catholic 49%, Other Christian 41%, Protestant 38%, Mormon 24%. White Evangelical 18%, not white Evangelical 58%. Never attends church 63%, attends church a few times a year 52%, attends church monthly 45%, attends church a few times a month 45%, attends church weekly or more 37%.

Put another way the religious breakdown of Biden voters (using the Votecast numbers) is:

White Evangelical 8%

Mormon <1%

Catholic 22%

Other Christian 26%

Non-Christian 45%

The religious breakdown of Trump voters (using Votecast numbers) is:

White Evangelical 39%

Mormon 2%

Catholic 23%

Other Christian 16%

Non-Christian 20%

Gender, Sexual Orientation and Marital Status

The CNN exit polls show Biden support at 45% for men and 57% for women. 

The Votecast exit polls show Biden support at 46% for men and 55% for women (the less than 1% of Votecast respondents who didn't identify as men or women had 70% support for Biden).

The CNN exit polls shows Biden support at 64% for LGBT voters (7% of the total) and 51% for non-LGBT voters.

CNN reports Biden support by marital status as:

Married (56% of voters) 46%

Married Men (59% of male voters) 44%

Married Women (53% of female voters) 47%

Unmarried (44% of voters) 58%

Unmarried Men (41% of male voters) 52%

Unmarried Women (47% of female voters) 63%

Votecast reports Biden support by gender and marital status (which interacts with age and race as well) as:

Married men (28% of voters) 42%

Married women (28% of voters) 47%

Unmarried men (18% of voters) 52%

Unmarried women (26% of voters) 62%

Age

The Votecast (CNN) breakdown of Biden support by age was:

18-29 61% (60%) 

30-44 54% (52%)

45-64 48% (49%)

65+    48% (47%)

CNN also had a different age breakdown with Biden support as follows:

18-24  65%

24-29  54%

30-39  51%

40-49  54%

50-64  47%

65+     47%

Household Income

Votecast showed very little difference in Biden support by household income with Biden support as follows:

Under $50,000 51%

$50,000 to $99,999 48%

$100,000 or more 53%

Age, gender, race, education and region all mattered much more in the Votecast results.

But, the CNN exit poll showed Biden and Trump support by income as follows:

Under $30,000 54-46%

$30,000 to $49,999 56-43%

$50,000 to $99,999 57-42%

$100,000 to $199,999 41-58%

$200,000 or more 44-44%

This shows a significant difference between pro-Biden lower and middle income voters and higher income pro-Trump voters, with results that flatly contradict the Votecast data.

By Region, Race and Religion

The CNN results are broken out by region. In that polling, by region and religion and race, Biden support was as follows:

                East-Midwest-South-West

White 46%-41%-32%-50%

Black 94%-81%-88%-77%

Latino 78%-67%-55%-67%

Other 75%-73%-28%-47% (tie).

Asian 63%-NA-NA-63%

It isn't clear to me what populations are driving the 4% of people in "other race" category in the CNN polls, particularly because it is so disparate regionally (even more so than white voters). The "other race" category is 4% of voters in the East, 5% in the Midwest, 3% in the South, and 5% in the West.

"Other race" might include a significant share of U.S. born Mestizo or Mulatto people rightly characterized as Hispanic who don't have a self-identification as "Latino", perhaps because they are assimilated into U.S. culture and are not fluent in the Spanish language.

Another possibility (not mutually exclusive), suggested by the unexpected conservative lean of Native Americans, Native Hawaiians and Pacific Islanders in the Votecast results, is that some people who identify as "American" described themselves as "Native American" or "Native Hawaiian" on the theory that they were born in the U.S. or in Hawaii, respectively, particularly in the South and West, while people in the East and Midwest who identify as "other race" may be mostly mixed race, South Asian or Middle Eastern. Per Wikipedia:

Pacific Islander Americans make up 0.5% of the U.S. population including those with partial Pacific Islander ancestry, enumerating about 1.4 million people. The largest ethnic subgroups of Pacific Islander Americans are Native Hawaiians, Samoans, Chamorros, Fijians, Palauans and Tongans.

There are significant Pacific Islander populations in American Samoa, Guam and the Mariana Islands, but none of those places have residents who vote in the Presidential general election. Native Hawaiians wherever they reside in the U.S., make up 43% of the total number of Pacific Islanders in the U.S., with a large share of the two largest runner up sub-categories of Pacific Islanders (Samoans 15.1%, and Chamorro 12.2%) living outside any U.S. state and outside the District of Columbia.

As of the 2010 census there were 1,164,575 Pacific Islanders who lived in a U.S. state or in the District of Columbia. There were 358,951 in Hawaii (26% of the population), 37,994 in Utah (1.9%), 11,360 in Alaska (1.5%), 320,036 in California (1.4%), 35,435 in Nevada (1.1%), 73,213 in Washington State (1.0%), 26,936 in Oregon (0.7%) and 28,431 in Arizona (0.4%) and smaller percentages of the population of every other state (including all states in the East, the Midwest and the South).

Meanwhile, about 0.9% of Americans are Native American or Alaska Natives. This is highest in Alaska (14.8%), New Mexico (9.4%), South Dakota (8.8%), Oklahoma (8.6%), Montana (6.3%), North Dakota (5.4%), Arizona (4.6%), Wyoming (2.4%), Washington  State (1.5%), Idaho (1.4%), Oregon (1.4%), North Carolina (1.3%), Nebraska (1.2%), Nevada (1.2%), Utah (1.2%), Colorado (1.1%), Minnesota (1.1%), California (1.0%), Kansas (1.0%) and Wisconsin (1.0%). The only states with a Native American population of more than 2% included in the CNN survey were Montana and Arizona.

The percentage of Native Americans, Alaska Natives, Native Hawaiians and Pacific Islanders who live outside the West is quite small.

                East-Midwest-South-West

Protestant/Other Christian 57%-37%-32%-40%

Catholic 43%-48%-58%-62%

Other NA-NA-70%-NA

None 60%-66%-54%-75%

The regional differences in Catholic support for Biden probably mostly reflect the relative proportions of white and Hispanic Catholics in these regions. Catholics in the East and Midwest are much more often likely to be white than in the South and the West.

It likewise isn't clear to me what religious self-identifications are driving the "other religion" category that accounts for 8% of respondents in the CNN polls (or the 8% of Votecast poll respondents who are "something else" religiously, either). This may be mostly people who identify as "spiritual but not religious" because their Biden preferences are fairly high.

According to Pew:

[T]he share of U.S. adults who identify with non-Christian faiths has ticked up slightly, from 5% in 2009 to 7% today. This includes a steady 2% of Americans who are Jewish, along with 1% who are Muslim, 1% who are Buddhist, 1% who are Hindu, and 3% who identify with other faiths (including, for example, people who say they abide by their own personal religious beliefs and people who describe themselves as “spiritual”).

Latino voters are 50% Catholic, 25% Protestant/Other Christian, 6% Other Religious Affiliation and 20% No Religious Affiliation in the CNN poll.

White voters are 25% Catholic, 46% Protestant/Other Christian, 3% Jewish, 6% Other Religious Affiliation and 20% No Religious Affiliation in the CNN poll.

Black voters are 10% Catholic, 56% Protestant/Other Christian, 3% Jewish, 14% Other Religious Affiliation, and 16% No Religious Affiliation in the CNN poll. The Jewish percentage should be zero, but 3% seems too high, although this could be simply a statistical fluke for a population that is 0.39% of the total sample of about 15,590 voters (i.e. about 61 black Jewish respondents, who may be more motivated to respond to an exit poll than the average black voter).

Voters in the East are 30% Catholic, 31% Protestant/Other Christian, 3% Jewish, 11% Other Religious Affiliation, and 24% No Religious Affiliation (a combined 38% non-Christian) in the CNN poll.

Voters in the Midwest are 28% Catholic, 47% Protestant/Other Christian, 2% Jewish, 8% Other Religious Affiliation, and 15% No Religious Affiliation (a combined 25% non-Christian) in the CNN poll.

Voters in the South are 19% Catholic, 56% Protestant/Other Christian, 2% Jewish, 8% Other Religious Affiliation, and 15% No Religious Affiliation  (a combined 25% non-Christian) in the CNN poll.

Voters in the West are 25% Catholic, 33% Protestant/Other Christian, 2% Jewish, 6% Other Religious Affiliation, and 35% No Religious Affiliation (a combined 43% non-Christian) in the CNN poll.

The percentage of voters who are Jewish also seems too uniform nationally.

State Variation By Race and Religion

The results are broken down state by state for the 24 states available (of which there are 18 with a Latino vote breakdown, 18 with a black vote breakdown, and 4 with an Asian vote breakdown), below the fold.

Black support for Biden was greater than Biden support in any other racial category in every state with a breakdown available. 

Asian support for Biden and Latino support for Biden was similar in all states where both data points were available.

Both Asian support for Biden and Latino support for Biden was greater than White support for Biden in every state where these data points were available.

White Support By State And By Religion

White support for Biden (in the 24 states surveyed) ranged from a low of 21% to Alabama to a high of 63% in Washington State, a 42 percentage point spread. Support for Biden was three times as great in the state with the most support as it was in the state with least support.

Broken down by religion, white support for Biden was highest among religiously affiliated non-Jewish, non-Christians at 65%, then no religious whites at 61%, then Catholics 44%, and then Protestants (and other non-Catholic Christians) at 27%. Among all born again or Evangelical white Christians it was 24% and among all whites who were not born again or Evangelical Christians it was 56%. 

Education explains some of the variation with 51% of college educated whites supporting Biden and 32% of whites who are not college educated supporting Biden, but this comes with an overlay of regional culture and religion that also partially drives this difference.

For example, 34% of white voters with college degrees with born again or Evangelical Christians, while 46% of white voters without college degrees were born again or Evangelical Christians.

Black Support By State and By Religion

Black support for Biden (in the 18 states with statistically significant samples) ranged from a low of 71% in Kentucky to a high of 94% in New York State, a 23 percentage point spread. Support for Biden was 32% greater in the state with most support from the state with least support.

Support for Trump among black voters was highest in states with quite small black populations (on a percentage basis). 

There was no break down by religion for black voters other than black Protestants/Other Christians which was identical to black voters as a whole. 

Latino Support By State And By Religion

Latino support for Biden (in the 18 states with statistically significant samples) ranged from a low of 53% in Florida to a high of 75% in New York State, a 22 percentage point spread. Support for Biden was 42% greater in the state with most support than in the state with least support. 

Broken down by religion, Latino support for Biden was highest among Catholics 71%, then those with no religious affiliation 55%, and then Protestants 51%.

Asian Support By State And By Religion

Asian support for Biden (in the 4 states with statistically significant samples) ranged from a low of 60% in Virginia to 76% in California, a 16 percentage point spread. Support for Biden was 27% greater in the state with most support from the state with least support. 

There was no break down by religion for Asian voters. 

Posting Rates In 2020

I am on track to have made about 8 posts per week in 2020 between this blog and its sister blog, Dispatches From Turtle Island. These are not record highs, but also making both blogs very active at a time when the solo blog format is becoming less common.

I have also added a new "page" to this blog regarding my citation and quotation policies today, which is linked in the sidebar.

Transliteration, Scripts and Orthography

Something of a scattered rant here.

Transliteration and Orthography

Transliteration is the process of writing a word not normally written with the Roman alphabet into Roman alphabet letters for people who don't understand a language's main script. There are consistent systems of doing so from one particular language, but this can lead to weird orthography.

Why are transliteration systems for various languages into the Roman alphabet so often at odds with intuitive pronunciation? 

Some examples:

* Why does transliteration of Maori transliterate an "f"-like sound as "wh"?

* Why does transliteration of Vietnamese transliterate a "wh"-like sound as "ng"?

* The revised Romanization of Korean adopted in 2000 by the South Korean ministry of culture is much further from a natural pronunciation of the transliterated words the way that they sound in Korean than the previous McCune–Reischauer system (which admittedly had too many accent type marks for the Internet and wasn't perfect itself). The new system is unambiguous leading instead to consistent error by the uninitiated instead of allowing people to sometimes get it right, especially if they've heard the word before but not seen it written.

* Why does the letter "X" used in so many disparate ways in transliterations, having different sounds, for example, in words derived from Greek, Chinese and Aztec?

It would be nice to have a simple reference of how non-intuitive transliterations work in all of the world's major languages in one place.

It would also be nice to have a list of what phonemes are absent in American English that are present in other major languages, and what phonemes are present in American English that are absent in particular other major languages.

Scripts

Why does Japanese of all language have three phonetic scripts (two based upon syllables that basically code every possible syllable in two different ways, and a Romanization)?

Japanese has a fairly small set of phonemes, pronounced in a consistent way, with few complexities such as a vowel later in a word influencing how a vowel earlier in the word is pronounced, few silent vowels and consonants, few vowels that are implied but not written, very few consonant clusters, little use of tone other than a pitch-accent system, etc. It could totally manage with a short and simple alphabet along the line of Hangul in Korean whose phonology is similarly straight forward.

The U.S. Military Sucks At R&D

The U.S. military and its contractors have a system that can develop extremely high technology systems. But its system also has deep flaws. 

For example, it took the U.S. military a year to develop an ordinary face mask (and this is fast by Army standards where 18-24 months for a simple innovation is the norm):

It’s not clear why the U.S. Army, the most powerful fighting force in the world, required nearly a year to develop a mask that would have taken the civilian sector mere days—if not hours—to develop. The only special features the covering has that civilian masks lack is the use of the OCP pattern and a military-style initialism (CCFC). . . 

The Army's struggle to develop a face mask recalls the service's similar efforts to buy a new handgun. It took the Army nine years to field a replacement for the M9 handgun. In 2016, the Chief of Staff of the U.S. Army, Gen. Mark Milley, told an audience: 

"We're not figuring out the next lunar landing. This is a pistol. Two years to test? At $17 million? You give me $17 million on a credit card, and I'll call Cabela's tonight, and I'll outfit every soldier, sailor, airman and Marine with a pistol for $17 million. And I'll get a discount on a bulk buy." . . .

The pistol program, which Army Times reports started in 2008, ended in 2017 with the selection of the Sig Sauer-made M17 handgun.

One of the reasons that Americans don't trust government is that we are so bad at carrying out governmental functions compared to many other countries.

The quote with the quote, by the way, is hyperbole. Even at a dirt cheap $300 per handgun, it would cost about $500 million to buy a pistol for every active duty service member, and no one would do it for less than $14 per pistol, as the general suggested rhetorically.

Internal Migration And Partisan Politics In The U.S.

 

 

The shift is driven by demographics—what demographers call "generational replacement," urbanization, and increasingly, the migration of blue state residents to red states.

From here. 

The source of the change appears to be apt and correct, although shifts in the character of the Republican party, which is shedding college educated whites and picking up whites without college educations, and the ongoing process of deindustrialization which drives both migration and changing political attitudes, also play a part.

Some notable details:

* Utah and Idaho have trended blue consistently from 2000-2020. These may be very conservative states due to Mormon influence, but the Mormon idea bubble is different from the Evangelical idea bubble, there is residual New England culture embedding in Mormon culture, and Utah is a lot more urban and educated than the vast majority of red states.

* Texas, Nebraska and Kansas have trended blue consistently from 2000-2020. These are also definitely red states, but Texas needs to shift by only about six percentage points to flip. Kansas is more educated than most other red states. 

* Virginia, Georgia and North Carolina are all trending blue, driven by metro DC, Atlanta, and the research triangle respectively.

* The Midwest, Deep South and Appalachia are trending red. So is Florida.

The Implications Of Better Electric Car Batteries

Sometime in the vicinity of the years 2024-2026, solid state batteries for electric cars will be widely available on a mass produced basis. This will dramatically improve the time required to charge the batteries, the lifetime of the batteries, their range, their energy density, and their safety. 

Electric cars already cost about half as much to maintain over the life of car and the electricity cost of charging them per mile traveled is far below the price of gasoline and diesel fuel. The maintenance cost will fall with solid state electric car batteries.

It is likely that this boost in performance and the increased scale of electric car production that results will bring down the total vehicle life cost of an electric cars to below that of a conventional internal combustion engine car, even on an unsubsidized basis. This critical tipping point will probably be reached sooner in Japan, Korea and Europe, where gasoline and diesel are about twice as expensive in the United States, before it is in the United States.

Once this tipping point is reached, in a decade or so, perhaps by 2035, internal combustion vehicles will become antiques and niche vehicles, with the vast majority of ground transportation vehicles powered by electricity, which will be increasingly green as coal fire power plants are replaced by wind, solar and tidal power, by natural gas, and perhaps by renewed growth in nuclear power generation capacity, the technology for which has been teed up for a while now.

Since internal combustion engines in vehicles are the predominant source of demand for petroleum, once this tipping point is reached, demand for petroleum will plummet, making high cost sources of petroleum, like fracking wells, much less economic than they are today.

Petroleum, since it is relatively easily transported over long distances in oil tankers, trades in a global market, and so the reduction in demand, even if initially limited to places like Europe and Japan and Korea, will impact the entire world.

Economies that rely significantly on oil production like Alaska, Oklahoma, Texas, Colorado, Venezuela, Saudi Arabia, the United Arab Emirates, Kuwait, Nigeria and Brunei will all be hard hit. Moreover, within those economies, the invariable conservative political interests that rely on oil wealth will see their power wane and the expense of the tech and commercial economies. Peak oil will become irrelevant, resulting not in extreme prices but in moderately low prices as demand falls dramatically, while supply falls less rapidly. 

The simultaneous continued deep decline in demand for coal which will be ongoing will result in continued economic malaise and depopulation in the two main states where coal production is significant, West Virginia and Wyoming, where the residual coal industry props up even more conservative political interests.

Phenomena like Islamic fundamentalism and monarchism, which are financed largely with oil wealth, despite the fact that it is ill adapted to a modern commercial economy, will lose the win in their sails, and will probably be replaced by some more practical iteration of Islam and by constitutional monarchies or republican revolutions in what short order transition from being affluent countries to struggling developing countries.

This transition away from coal and petroleum (although natural gas demand should remain healthy) will dramatically arrest new man made contributions to global warming and will greatly reduce air pollution. Even Los Angeles will have clean air.

Less strikingly, this will sink demand for mechanics and mechanical engineers, and increase demand for electricians and electrical engineers.

At the neighborhood business level, gas stations that fail to transition to electric charging stations (and it isn't clear that the new electric charging infrastructure will be a good fit for old gas stations anyway), will go out of business in droves. A few will linger with legacy gasoline and diesel demand and convenience store businesses, but many will not.

Golden Kamuy

I've been watching the anime, Golden Kamuy, on Crunchyroll lately. 

It is a period piece set shortly after the Russo-Japanese war of 1904-1905, in Hokkaido, the northernmost main island of Japan, featuring a veteran of the war and an orphaned Ainu girl who has just reached marriageable age but is too much of a tom boy to be attractive to Ainu men, who team up to find a stash of gold stolen from the Ainu people in a robbery that killed her father and to avenge her father's death. The location of the gold was tattooed in a treasure map distributed amongst multiple escaped prisoners and the remnants of a disgraced Army division or groups of prisoners are also hot on the trail.

I've researched the factual points touched upon by the story and while it takes liberty with our reality, the alternative history isn't too far off.

The disgraced Army division's efforts closely mirrors that of the people who briefly formed the Ezo Republic, although the timing is about forty years after the real events that it echos. The scenes from the Russo-Japanese war (in flashbacks) likewise convey well the core and defining realities of that conflict such as the fact, that would be reduplicated ad nauseam in World War I, a decade later, that a well positioned small force with machine guns is a very tough match for a large force of infantry armed only with bolt action rifles and bayonets, and the acid bath of class tension that resulted from repeated orders to continue futile or extremely costly infantry charges despite this fact.

A relict population of Ezo wolves, that probably went extinct in the 1880s or 1890s, is prominent in the story, but given that the subspecies of Hokkaido wolves which was much more closely related to the gray wolves of North America than to Asian wolves was believed to be extinct already in the story itself, this isn't such a great departure.

The notion of Hokkaido as a center of gold mining, mostly panning rivers for gold, is authentic and its remains the center of gold mining in Japan. 

Everything depicted about Ainu culture and religion in the anime likewise closely matches official sources about the real world, including the encouragement that the young Ainu woman's grandmother gives to her to marry her ethnically Japanese partner.

And, of course, the landscapes and urban backgrounds are clearly based upon contemporaneous images of the real thing.

On the whole, the choices made to balance the needs of an exciting story and the goal of conveying the gist and context of a specific historical period in a particular place that it brings to life, are very well done. I would recommend it.

The Final 2020 Election Results

The Final Presidential Election Results

The electoral vote margin in favor of Biden was 64 electoral votes out of 538 electoral votes. Biden won in 25 states and the District of Columbia. Trump won in 25 states.

The popular vote margin in favor of Biden was 4.5 percentage points. Wisconsin was the marginal state and favored Biden by 0.7 percentage points. 

These Results Are Really Final

The results below reflect the Presidential vote totals and electoral college votes certified on the December 8, 2020 deadline for doing so in all 50 states and the District of Columbia (including the three Congressional Districts of Nebraska and two Congressional Districts of Maine that cast separate electoral votes with the state winner taking an additional two electoral votes). 

All meaningful possibilities of a court challenge to any of the results have been exhausted. 

Even an unprecedentedly high number of unfaithful electors (never more than a low single digit number in past elections) would not sway the actual casting of the Electoral College vote on Monday, December 14, 2020.

And, since Democrats control one of the two houses of Congress, it is effectively impossible for Congress to vote to refuse to acknowledge the electoral vote count due to disputes raised by Republicans to the outcome.

The Electoral College Systemically Favors The GOP

While the Electoral College made this election a close one, the popular vote in the 2020 Presidential election wasn't close. The systemic bias of the electoral college in favor of the Republican relative to the popular vote is about 3.8 percentage points of the popular vote. A Democrat needs more than a 3.8 percentage point lead in the popular vote to win the Electoral College.

Also, despite the fact that Democrats control the majority of the seats in the U.S. House, if the election had been an electoral college tie of 269-269 electoral votes, the vote of the U.S. House by state delegations in that scenario would have re-elected Trump. So, Biden had to win 270 electoral votes to win, while Trump only had to win 269 electoral votes to win.

Third-Parties Probably Didn't Change The Outcome

Third parties won 1.9% of the popular vote, which was less than Biden's popular vote margin of victory. If the Libertarian party had not run a candidate in this election, the outcome would have been much closer, but Biden would have very likely still won the Presidential election in the end.

Third-party voting could conceivably have flipped Arizona and Georgia to Biden from Trump (which wouldn't have changed the overall electoral college outcome) where Biden's margin of victory was razor thin and the third-party candidates on the ballot were more conservative than liberal leaning.

In Georgia, Biden won the state by a margin of 11,779 votes (0.2% of the total), there were no write-in votes, and the only third-party candidate on the ballot was Libertarian Jo Jorgensen. who secured 62,229 votes (1.2% of the total). If all of the Libertarian voters had voted for Biden or Trump instead in a two candidate race, Trump would have needed 59.5% of the Libertarian vote to win the state, which is well within the range of plausibility.

In Arizona, Biden won the state by a margin of 10,457 votes (0.3% of the total), there were 1,928 write-in votes, and the only third-party candidate on the ballot was Libertarian Jo Jorgensen. who secured 51,465 votes (1.5% of the total). If all of the Libertarian voters had voted for Biden or Trump instead in a two candidate race, Trump would have needed 60.2% of the Libertarian vote to win the state, which is again well within the range of plausibility.

But third-party voting probably didn't ultimately change the outcome in  the marginal state of Wisconsin relative to what it would have been in a two candidate race (although a lack of third-parties would have probably led to a closer race). Biden's margin of victory in Wisconsin was 20,740 votes. The final outcome in the Wisconsin Presidential race was:

Joe Biden 1,630,930 votes (49.5%)

Donald Trump 1,610,190 votes (48.8%)

Jo Jorgensen (Libertarian) 38,493 votes (1.2%)

Write-ins 8,087 votes (0.2%)

Brian Carroll (Amer. Solidarity Party) 5,259 (0.2%)  

The Libertarian party is socially liberal and economically conservative. The American Solidarity Party is a center-right Christian Democratic Party that is more liberal than the current Republican Party. Write-in candidates aren't consistently left or right leaning and reflect a group of voters who would have been particularly unlikely to vote in the Presidential race at all if it had been only a two candidate race.

If all of the people who voted third-party in the Presidential election voted in that race if no third-parties had been on the ballot, Trump would have needed about 70.6% of third-party voters to win in Wisconsin (or 73.6% of the vote cast for candidates who were not write-in candidates). Given that both of the main third-party candidates in the Wisconsin race are between Biden and Trump politically, and that any conservative leaning voters who voted for these candidates did so, in part, out of a serious distaste for voting for Trump even though they would usually vote for a Republican in a two party race, this would be a tall order, although Trump might get more than 50% of the third-party vote.

Also prior studies have suggested that about half of third-party voters wouldn't have voted in the Presidential election at all if a third-party candidate wasn't on the ballot, in which case Trump would have needed 91.2% of the third-party vote to win in Wisconsin, something he almost surely could not have done. But, even if so smaller but significant share of Libertarian and American Solidarity Party voters didn't vote for either candidate in a two candidate race, the percentage of the remaining third-party voters who would have to vote for Trump for him to win would be even greater than the 70% needed if all third-party voters still voted, or the 73% needed if all Libertarian and American Solidarity Party voters voted for either Biden or Trump.

Third-parties definitely didn't have a decisive impact on the outcome in any other states. Third-party candidates combined received fewer votes than the margin of victory for Biden in every other state except Pennsylvania.

In Pennsylvania, Biden won a plurality of the vote with a 1.2 percentage point lead, but was 2,705 votes short of an outright majority of the Presidential votes cast. There were 79,397 votes cast for Libertarian Jo Jorgensen (1.1% of the vote) and there were 7,672 votes (0.1% of the vote) cast for write-in candidates. Even if every single one of those voters cast a vote in a two candidate race (which is unlikely), Trump would have needed to win 96.9% of the third-party vote (which would be all of the Libertarian vote and about 65% of the write-in vote in addition) to win Pennsylvania, which is not a plausible possibility.

So, to restate my initial conclusion, third-party candidates probably did increase Biden's margin of victory and may have even caused him to win Georgia and Arizona, but probably would not have flipped the entire Presidential race to Trump in a two candidate race.

Where Was The Third-Party Vote High And Low?

Another interesting point regarding the third-party vote is that voters appear to understand to some extent that a third-party candidate acts as a spoiler and act accordingly. 

Generally speaking, the third-party vote was highest in states which were (or were perceived as being prior to the election) safe for one Presidential candidate or the other, and were lowest in states which were (or were perceived as being prior to the election) battleground states.

The third-party vote was more than two and a half percentage points in Arkansas (4.8%), Alaska (4.4%), and Utah (4.3%), Wyoming (3.5%), Vermont (3.2%), Washington (3.2%), Oregon (3.1%), North Dakota (3.1%), Idaho (3.1%), Montana (2.9%), Maine (2.9%), Colorado (2.7%) and South Dakota (2.6%).

The third-party vote was lowest in Florida (0.9%), the only state where it was below one percentage point and a state with a history of very close results that had been (inaccurately) predicted to be very close in 2020. Every state that was close or predicted to be close had a third-party vote of 2.2% or less.

In addition to being higher in states perceived as safe and lower in states perceived as battleground states, dissatisfaction with the safe candidate in a state can also help explain the magnitude of the third-party vote. 

The Biden supporting states with a third-party vote of more than two and half percent were states, it is plausible to think, where the third-party vote was driven by Sanders progressives dissatisfied with Biden because he is perceived as a political moderate. 

At least some of the Trump supporting states with a third-party vote more than two and a half percent were states, it is plausible to think, where solidly conservative states that were dissatisfied with Trump as an individual candidate.

A Swing State Analysis

Only nine jurisdictions had a margin for the winning candidate of seven or fewer percentage points. From most pro-Biden to most pro-Trump, those states were: Michigan, Nevada, Pennsylvania, Wisconsin, Arizona and Georgia, all won by Biden (and all, except Nevada, not won by Hillary Clinton in 2016), and North Carolina, Florida and Texas, won by Trump. Trump did not win a single state that he did not win in 2016.

The margin of victory for the winner was 1.3 percentage points or less in Pennsylvania, Wisconsin, Arizona, Georgia and North Carolina.

Biden had a decisive lead in 20 jurisdictions, in addition to the 2 he won by a moderate margin and 4 he won by small margins. Trump had a decisive lead in 23 jurisdictions, in addition to 2 he won by a moderate margin and1 he won by  a small margin.

Of the swing states, demographic change and long term trends probably favor Democrats in 2024 relative to this year's election in Nevada, Arizona, North Carolina, Florida and Texas, although not necessarily enough for Democrats to win in Florida or Texas through demographic trends alone. 

The long term trend lines probably hurt Democrats in Michigan, Pennsylvania and Wisconsin in 2024, which have grown more conservative in recent years. They have grown more conservative because their rust belt economies have pushed liberal voters to migrate away from the state, and because voters that have remained to become more conservative due to their declining economic fortunes.

Detailed Results

The state by state results are as follows:

From here and here and here and here.

Updated polling error analysis

The pre-election polling was inaccurate in a year where there was no good reason for it to be, and leaned strongly and consistently in favor of Biden relative to the actual results known in 20/20 hindsight. 

As I've explained elsewhere, the most likely cause was widespread depressed participation in election polling by Trump supporters as a result of their social distrust in a manner that merely adjusting polling results by the demographics of the respondents wasn't sufficient to cure. In hindsight, this was probably also the main problem in the 2016 polling errors which were similar to, but not quite as large, as in 2020.

In the chart above, the column marked 538 Net Biden reflects the expected margin in percentage points of Biden over Trump according to final polling averages compiled by the Five Thirty-Eight website on the eve of the November 3, 2020 election, less the actual margin in percentage points of Biden over Trump.

The pre-election polling averages got only two states wrong (both of which were won by Trump): North Carolina and Florida. But several states that the polling projected that Trump would just barely win were easily won by Trump (Texas, Ohio and Iowa). And several states that the polling projected that Biden would easily win were won by Biden by much smaller margins than predicted (Wisconsin, Pennsylvania, Nevada and Michigan).

Just 4 jurisdictions out of 51 had a pro-Trump bias in pre-election polling averages: District of Columbia, Maryland, Colorado and Utah. The errors in the first three jurisdictions were all small. The error was biggest in Utah and  matched by one of the largest third-party share of the Presidential vote in the nation (4.3%), which is notable because a large third-party vote tends to make polling predictions less accurate. All of the states with a pro-Trump lean in the polls were jurisdictions which were safe for the winning candidate. The other 47 out of 51 jurisdictions had a pro-Biden polling bias. The average pro-Biden bias in the Presidential polling was 4.24 percentage points. 

The average overall error in polling was 4.54 percentage points. If the polls were adjusted for the average pro-Biden bias (bringing the average bias in the polls to zero), the average overall error in polling would have been 2.67 percentage points, which is very close to what would be normally be expected from random sampling error, corroborating the inference that the sole reason that the otherwise high quality polling averages were off was a failure to correct for lower response rates from Trump supporters than from Biden supporters, all other things being equal.

There were polling errors of four or more percentage points (all in favor of Biden) in 27 out of 51 jurisdictions. A standard deviation in a polling average should be well under four percentage points, so there shouldn't have been less than 16 polling errors of four or more percentage points and they should have been roughly evenly split between favoring Biden and favoring Trump, if the sampling were random.

The U.S. House Races

In the end, Democrats won 222 seats in the U.S. House (more than the 218 seats needed for a majority) and Republicans won 213 seats. 

The closest races were the Second Congressional District of Iowa which the Republican won by just 6 votes and the Twenty-Second Congressional District in New York which the Republican won by just 12 votes (subject to a judicially ordered recount which is underway).

This is 13 seats less than the 235 seats in the House that Democrats held in the U.S. House after the 2018 midterms and the special election in North Carolina held due to Republican voting fraud in the original race.

Democrats in the House will be beholden to the most centrist 2% of their members because just five net defections from the party line are sufficient to cause them to lose a majority on any particular legislation.

The U.S. Senate Races

The Democrats have secured 48 seats in the next session of the U.S. Senate, and the Republicans have 50 seats. Two more U.S. Senate seats in Georgia will be decided based upon a January 5, 2021 runoff election, because the Republicans in each of those races won pluralities, but not a majority of the November 3, 2020 election vote.

The Democratic candidates are narrowly leading in the polls in both of the Georgia Senate races. Democrat Ossoff has a 1.0 percentage point lead over Republican Perdue. Democrat Warnock has a 1.6 percentage point lead of Republican Loeffler.

The pro-Biden bias in the Presidential polling in Georgia in 2020 was an unusually low 0.7 percentage points. The Democratic bias in the polling in the Georgia regular election race between Ossoff and Perdue was 1.7 percentage points. The Democratic bias in the polling in the Georgia special election race which is now between Warnock and Loeffler was 6.0 percentage points, but that is in part because is was a hard to accurately poll many candidate race without a primary.

Given the polling biases seen this year in Georgia's general election races, both U.S. Senate runoff elections look like coin flips in which the Democrats are underdogs, but only slightly so.

If the Democrats win both U.S. Senate races in Georgia, then the Senate will be split 50-50 and Democrat Kamala Harris, as Vice President will resolve any tie votes, giving Democrats control of the Senate by the narrowest of margins. If Democrats lose one or both of the runoff elections, then Republicans will control the Senate for at least the next two years.

Even if the Democrats secure control of the Senate, even one net defection from the party line will cost them a majority on any particular issue, and some its members, like Democrat Joe Manchin from West Virginia (the second most pro-Trump state in the U.S. after Wyoming in 2020), are quite conservative relative to the rest of the Democratic caucus in the Senate.

On the other hand, even if Republicans secure a 51 or 52 seat majority, with the Vice President's tie casting vote in the hands of the Democrats, it will take only one or two net defections from Republicans for Democrats to win particular votes on the merits, and Republicans can't overcome Democratic filibusters of ordinary legislation without exercising the "nuclear option" which doesn't make much sense for them to do when any legislation that passes the Senate over a Democratic filibuster attempt could still be scuttled in the Democratic party controlled U.S. House or with a veto from President Biden.

Instead, if Republicans manage to secure a U.S. Senate majority, Mitch McConnell will no doubt continue his long standing practice of using the Senate to prevent Democratic Party backed legislation from the House from becoming law and attempting to block or delay President Biden's judicial and executive branch nominations.