Some observations on an extended election night:

1. This was not a good night for conventional polling. My review in Tuesday’s Wall Street Journal of a book on the history of “polling failures” took perhaps too positive a view of contemporary polling.

I find it remarkable that polling has been as accurate as it has been in a country where the completion rate for pollsters’ contacts is below 10% — but it got worse this week.

The Real Clear Politics average of recent polls showed Joe Biden with more than 51% of the popular vote and Donald Trump with 44%. As this is written, Biden has 50% of the tabulated national popular vote, which will probably rise as California’s data comes dribbling in, but Donald Trump has 48%.

So, the current 1.9% Biden plurality is far lower than the polls’ 7.2% Biden plurality. The results in target states seem to have been off, as well. The polls had Biden up 0.9% in Florida, far different from Trump’s 3.4% victory margin there.

I have been dubious about the polling techniques of firms like Robert Cahaly’s Trafalgar, which produced numbers apparently closer to the election results so far than other pollsters in Florida, North Carolina, Pennsylvania, Michigan and Wisconsin. But in an opinion climate where mass media and corporate political correctness has many Americans unwilling to state their opinions, there may be something to say for Cahaly’s unorthodox methods — and something to say against the more standard polling technique.

There is, as my Washington Examiner colleague Byron York has argued, a hidden Trump vote. Most pollsters have not learned how to find it.

2. Donald Trump is running much better than almost everyone in the press and on Twitter expected. As I write, he has won or is ahead in the tabulated vote in 28 states with 280 electoral votes. He is clearly on his way to winning more popular votes than he did in 2016.

Democrats don’t seem likely to pick up the Senate majority that they seemed almost assured of gaining as recently as Tuesday morning. Republicans Corey Gardner and Martha McSally lost, as expected, in Colorado and Arizona, and Democrat Doug Jones lost, as expected, in Alabama.

But as I write, Susan Collins is running well ahead of Trump and has a significant lead in Maine. She is a rare example of a politician who can get voters to split their tickets.

Cal Cunningham, Chuck Schumer’s handpicked candidate in North Carolina, is trailing by 1.8%; I bet if Schumer if had known about Cunningham’s extramarital affairs, which were revealed and admitted during the campaign, the Senate minority leader would have picked another candidate.

In Iowa, Republican incumbent Joni Ernst, having trailed in polls for months, won by a 7% margin. This is even larger than the prediction shown in last weekend’s Des Moines Register poll, conducted by star pollster Ann Selzer, about which I wrote a column entitled “The Disconcerting Iowa poll: A Trump Surge?”.

Democrats’ hopes for a Senate majority now rest on Georgia’s two Senate seats.

It’s now looking unlikely that incumbent David Perdue will fail to get 50% of the vote and therefore, under state law, have to face a runoff with Democrat Jon Ossoff on Jan. 5; Perdue currently holds 50% of the vote with 98% of precincts reporting. There will definitely be a runoff for the state’s other seat between appointed incumbent Kelly Loeffler and Democrat Raphael Warnock. Old-timers may recall that in 1992, when Georgia voted for Bill Clinton, Republican Paul Coverdell trailed incumbent Wyche Fowler in the general election 49% to 48%, but came back to win the runoff, with a sharply reduced turnout, by 51% to 49%.

This year, with the Senate majority potentially at stake, turnout is likely to remain high.

Whatever the exact Senate result, it seems unlikely that the Senate will go along with the plans backed enthusiastically by leftist Democrats to pack the Supreme Court or to confer statehood on the District of Columbia and Puerto Rico.

3. Democrats have not done nearly as well as expected in House races. The Cook Political Report, with its fine record of assessing congressional elections, predicted that Democrats would increase their 232-seat majority in the House. Instead, at present, they’ve lost multiple seats and have gained only two, both in North Carolina, thanks to a favorable court redistricting decision.

One consequence of the House election results: Republicans went into the election with majorities in 26 state delegations. That’s important, because the Constitution mandates that if no presidential candidate has a majority in the Electoral College, the presidency is decided by the House, with each state casting one vote and a majority required to elect. So, if the Electoral College is divided 269-269, Republicans with 26 or more state delegations could elect the president.

4. This was, as our Washington Examiner editorial argues, a bad election for identity politics. It’s also been bad for the theory unfurled most famously in the 2004 book “The Emerging Democratic Majority” that increasing percentages of nonwhites would assure Democrats of a permanent natural majority. But there never was any assurance that the Americans classified (beginning with the 1970 census) as Hispanics would overwhelmingly identify as Democrats, just as Black Americans have for the last half-century; the contemporary and historic experiences of these groups are far different.

But Hispanic voters didn’t perform as expected, and neither did Black voters. National exit polls showed Trump winning the votes of 32% of Latinos and 36% of Latino men; it also showed Trump winning 12% of Black voters and 18% of Black men. The gender gap is apparently widening among Latino voters and Black voters.

Hispanics were particularly prominent in determining the results in Florida and Texas.

Miami-Dade County, which casts 11% of Florida’s votes, voted 63% to 34% for Hillary Clinton in 2016 but gave Joe Biden only a 53 to 46% margin in 2020. And Democrats lost two congressional seats anchored in Miami-Dade. Cuban Americans in particular, who were open to Democrats in the Obama elections and in 2016, were successfully wooed by Donald Trump as they had been by Florida Republicans Rick Scott and Ron DeSantis in the 2018 Senate and gubernatorial races (polls showed both behind, but they won; sound familiar?).

The Trump campaign ran catchy ads aimed at Florida Hispanics and reminded them that the Democrats stood for “socialism,” which people with backgrounds in Cuba, Venezuela and other Latin countries do not think is a good thing. In south Texas as well, Trump carried almost every county in the lower Rio Grande Valley, an area that has usually gone 2-1 Democratic.

There’s more to learn about these trends, but one thing that is obvious is that Democratic strategists and journalists who like to use the campus-invented term “Latinx” are out of touch with American Latinos, only 3% of whom identify with that term.

One more result worthy of notice: California’s overwhelmingly Democratic legislature put on the ballot a proposition repealing Proposition 209, passed in 1996, which banned racial discrimination by state and local governments, including in college and university admissions.

Some legislators with many Asian constituents complained, but the assumption was that it would pass easily. Not so. The current returns show this Proposition 16 rejected by a 56% to 44% margin. It is barely carrying Los Angeles County with 52% and the San Francisco Bay area with 53%, but is rejected by 64% in the rest of the state, where most California voters live.

Evidently something sticks in the craws of most Californians, like most Americans, when they are asked to authorize discriminating against their fellow citizens because of their race.