The popular vote on election night

It seems that it was in 2004 that networks began posting the popular vote as a regularly recurring feature on their graphics. (I am taking all the networks--ABC, NBC, CBS, Fox News, CNN, and MSNBC--as essentially equally reliable.) Times are Eastern.

2004

Bush 30,086,416 [51%], Kerry 28,166,463 [48%] [CNN] [before the 11 PM poll closings]
Bush 33,044,406 [51%], Kerry 31,205,991 [48%] [CNN] [after the 11 PM poll closings]
Bush 36,699,167 [51%], Kerry 34,772,694 [48%] [CNN]
Bush 38,507,196 [51%], Kerry 36,445,392 [48%] [CNN]
Bush 41,076,201 [51%], Kerry 38,572,829 [48%] [CNN]
Bush 43,072,551 [51%], Kerry 40,433,228 [48%] [CNN]
Bush 44,926,339 [51%], Kerry 42,106,133 [48%] [CNN]
Bush 46,578,994 [51%], Kerry 43,490,962 [48%] [CNN]
Bush 48,157,684 [51%], Kerry 44,812,454 [48%] [CNN]
Bush 49,200,728 [51%], Kerry 45,784,980 [48%] [CNN]
Bush 50,285,850 [51%], Kerry 46,889,609 [48%] [CNN]
Bush 51,414,770 [51%], Kerry 47,810,296 [48%] [CNN]
Bush 52,295,249 [51%], Kerry 48,707,861 [48%] [CNN]
Bush 53,191,518 [51%], Kerry 49,575,120 [48%] [CNN]
Bush 54,013,221 [51%], Kerry 50,269,645 [48%] [CNN]
Bush 54,526,433 [52%], Kerry 50,694,741 [48%] [CNN]
Bush 55,184,508 [51%], Kerry 51,439,604 [48%] [CNN]
Bush 55,679,528 [51%], Kerry 51,923,333 [48%] [CNN]
Bush 56,132,833 [51%], Kerry 52,322,710 [48%] [CNN]
Bush 56,566,858 [51%], Kerry 52,864,308 [48%] [CNN]
Bush 56,787,311 [51%], Kerry 53,076,006 [48%] [CNN]
Bush 57,104,549 [51%], Kerry 53,481,762 [48%] [CNN]
Bush 57,285,649 [51%], Kerry 53,633,698 [48%] [CNN]
Bush 57,429,517 [51%], Kerry 53,755,698 [48%] [CNN]

2008

Obama 34,715,218 [51%], McCain 32,953,932 [48%] [NBC, before 11 PM poll closings]
Obama 37,051,373 [51%], McCain 34,900,640 [48%] [NBC, shortly after 11 PM poll closings]
Obama 43,217,769 [51%], McCain 40,374,626 [48%] [NBC, during McCain's concession speech]
Obama 48,195,195 [51%], McCain 44,757,863 [48%] [NBC, shortly before Obama's victory speech]
Obama 50,927,735 [51%], McCain 47,128,139 [47%] [NBC, shortly after Obama's victory speech]
Obama 52,195,702 [51%], McCain 48,197,998 [47%] [NBC]
Obama 53,394,419 [51%], McCain 49,192,880 [47%] [NBC]
Obama 54,315,351 [52%], McCain 49,742,538 [47%] [NBC]
Obama 55,156,495 [52%], McCain 50,498,219 [47%] [NBC]
Obama 56,455,824 [52%], McCain 51,403,831 [47%] [NBC, 2:00 AM]
Obama 56,712,640 [52%], McCain 51,563,859 [47%] [NBC]

Obama's victory speech began 'shortly before midnight ET'.

2012

Romney 33,451,829 [51%], Obama 31,871,583 [48%] [CNN, 10:47 PM]
Romney 38,456,135 [50%], Obama 37,256,428 [49%] [CNN, 11:12 PM]
Romney 41,534,759 [50%], Obama 40,661,832 [49%] [CNN, 11:25 PM]
Romney 43,302,613 [50%], Obama 42,781,092 [49%] [CNN, 11:33 PM]
Romney 44,477,318 [49%], Obama 44,278,104 [49%] [CNN, 11:38 PM]
Romney 46,335,966 [49%], Obama 46,110,767 [49%] [CNN, 11:55 PM]
Romney 49,193,964 [49%], Obama 49,159,858 [49%] [CNN, 12:27 AM]
Romney 49,227,135 [49%], Obama 49,204,276 [49%] [CNN, 12:28 AM]
Romney 49,305,636 [49%], Obama 49,258,390 [49%] [CNN, 12:29 AM]
Obama 49,495,060 [49%], Romney 49,475,300 [49%] [CNN, 12:32 AM]
Obama 50,990,159 [49%], Romney 50,754,834 [49%] [CNN, 12:56 AM]
Obama 51,711,918 [49%], Romney 51,443,569 [49%] [CNN, 1:11 AM]
Obama 52,540,656 [49%], Romney 52,030,140 [49%] [CNN, 1:28 AM]
Obama 53,309,487 [50%], Romney 52,628,611 [49%] [CNN, 1:40 AM]
Obama 55,613,347 [50%], Romney 54,364,318 [49%] [CNN]
Obama 56,268,504 [50%], Romney 54,815,431 [49%] [CNN]

2016

Trump 36,107,482 [49.1%], Clinton 34,382,220 [46.7%] [Fox News, 10:35 PM]
Trump 43,038,272 [49.0%], Clinton 40,965,265 [46.6%] [Fox News, 11:14 PM]
Trump 46,307,263 [48.8%], Clinton 44,366,620 [46.7%] [Fox News, 11:28 PM]
Trump 48,861,636 [48.5%], Clinton 47,317,010 [47.0%] [Fox News, 11:47 PM]
Trump 50,960,682 [48.3%], Clinton 49,823,466 [47.2%] [Fox News, 12:07 AM]
Trump 51,740,830 [48.2%], Clinton 50,596,040 [47.2%] [Fox News, 12:19 AM]
Trump 52,051,634 [48.2%], Clinton 50,923,182 [47.2%] [Fox News, 12:23 AM]
Trump 54,344,525 [48.3%], Clinton 52,984,683 [47.1%] [Fox News, 1:04 AM]
Trump 55,441,249 [48.2%], Clinton 54,194,607 [47.1%] [Fox News, 1:34 AM]
Trump 55,611,240 [48.2%], Clinton 54,388,568 [47.1%] [Fox News, 1:39 AM]
Trump 55,756,736 [48.2%], Clinton 54,554,510 [47.1%] [Fox News, 1:45 AM]
Trump 56,516,620 [48.1%], Clinton 55,354,663 [47.1%] [Fox News, 2:16 AM]
Trump 56,620,258 [48.1%], Clinton 55,496,416 [47.2%] [Fox News, 2:19 AM]
Trump 56,820,164 [48.1%], Clinton 55,817,910 [47.2%] [Fox News, 2:31 AM]
Trump 56,918,173 [48.1%], Clinton 55,900,992 [47.2%] [Fox News, 2:36 AM]

Trump 54,676,327 [48.3%], Clinton 53,309,785 [47.1%] [CNN, 1:13 AM]
Trump 56,897,955 [48.1%], Clinton 55,877,015 [47.2%] [CNN, 2:33 AM]
Trump 57,093,682 [48.0%], Clinton 56,088,589 [47.2%] [CNN, 2:48 AM]
Trump 57,270,082 [48.0%], Clinton 56,339,307 [47.2%] [CNN, 3:04 AM]
Trump 57,514,921 [48.0%], Clinton 56,711,981 [47.3%] [CNN, 3:15 AM]
Trump 57,948,438 [47.8%], Clinton 57,373,655 [47.4%] [CNN]
Trump 58,111,692 [47.8%], Clinton 57,623,876 [47.4%] [CNN, a few minutes after 4:08 AM]
Trump 58,372,808 [47.7%], Clinton 58,046,092 [47.5%] [CNN, 4:58 AM]
Trump 58,693,565 [47.7%], Clinton 58,544,444 [47.5%] [CNN, 5:56 AM]
Trump 58,735,977 [47.6%], Clinton 58,645,995 [47.6%] [CNN, several minutes after 6:05 AM]
Clinton 59,059,121 [47.6%], Trump 58,935,231 [47.5%] [CNN, 7:19 AM]
Clinton 59,344,081 [47.7%], Trump 59,178,831 [47.5%] [CNN, 10:37 AM]
Clinton 59,390,851 [47.7%], Trump 59,215,097 [47.5%] [CNN, 11:08 AM]
Clinton 59,415,096 [47.7%], Trump 59,229,732 [47.5%] [CNN, 11:34 AM]
Clinton 59,916,416 [47.7%], Trump 59,690,923 [47.5%] [CNN, 10:13 PM, Wednesday]

2020

On election night 2020, Fox News was somewhat inconsistent in how it updated viewers on the popular vote. On at least three occasions around 1 AM on the East coast (as well as at least once before the 11 PM poll closings), it had a graphic feature that gave both the raw vote numbers as well as the percentages. Between 11 PM and about 12:45 AM, however, their regular popular vote graphic feature gave raw numbers only, and they seemed to switch back to this at some point after about 1:20 AM.

Trump 43,291,659 [50.6%], Biden 40,868,897 [47.8%] [Fox News, 10:47 PM]
Trump 47,420,592, Biden 45,164,855 [Fox News, 11:10 PM]
Trump 50,324,858, Biden 48,974,987 [Fox News, 11:20 PM]
Biden 55,100,764, Trump 53,767,696 [Fox News, 11:30 PM]
Biden 58,292,901, Trump 57,107,087 [Fox News, 12:00 AM]
Biden 62,105,415 [49.8%], Trump 60,543,470 [48.6%] [Fox News, 12:47 AM]
Biden 63,425,040 [49.8%], Trump 61,810,899 [48.6%] [Fox News, 1:17 AM]
Biden 64,201,544 [49.9%], Trump 62,495,745 [48.5%] [Fox News, 1:22 AM]
Biden 64,743,760, Trump 63,014,167 [Fox News, 1:35 AM]

On at least three occasions after 1:22 AM, the camera showed a close-up of the information wall on the Fox News set which showed the national popular vote percentages as well as raw vote:

Biden 65,217,824 [49.9%], Trump 63,461,934 [48.5%] [Fox News, 1:48 AM]
Biden 66,632,739 [49.8%], Trump 64,999,804 [48.6%] [Fox News, 2:48 AM]
Biden 66,786,710 [49.8%], Trump 65,140,157 [48.6%] [Fox News, 3:00 AM]

Fox News featured a popular vote graphic, including percentages, nearly continuously in its next-day coverage:

Biden 69,141,253 [50.0%], Trump 66,859,610 [48.4%] [Fox News, 9:00 AM, Nov. 4]
Biden 69,285,329 [50.0%], Trump 66,952,705 [48.3%] [Fox News]
Biden 69,472,830 [50.1%], Trump 67,061,150 [48.3%] [Fox News]
Biden 69,866,826 [50.1%], Trump 67,233,023 [48.2%] [Fox News]
Biden 69,935,569 [50.2%], Trump 67,254,057 [48.2%] [Fox News]
Biden 70,612,216 [50.2%], Trump 67,679,183 [48.1%] [Fox News]
Biden 70,672,465 [50.2%], Trump 67,703,927 [48.1%] [Fox News, 3:59 PM, Nov. 4]

As well as on November 5, after 9 AM:

Biden 72,144,077 [50.4%], Trump 68,671,555 [48.0%] [Fox News, 9:01 AM, Nov. 5]
Biden 72,297,195 [50.4%], Trump 68,797,809 [48.0%] [Fox News, 11:55 AM, Nov. 5]

As well as on more isolated occasions on Nov. 6:

Biden 73,488,248 [50.5%], Trump 69,622,407 [47.8%] [Fox News, 5:56 AM, Nov. 6]
Biden 73,523,733 [50.5%], Trump 69,635,874 [47.8%] [Fox News, 9:00 AM, Nov. 6]

And Nov. 7:

Biden 74,838,997 [50.6%], Trump 70,588,275 [47.7%] [Fox News, 11:02 AM, Nov. 7]

NBC's graphic feature consistently gave percentages as well as raw numbers, but I can't correlate their data with specific times:

Trump 34,129,558 [50.6%], Biden 32,393,290 [48.0%] [NBC]
Biden 58,499,374 [49.9%], Trump 56,909,694 [48.6%] [NBC]
Biden 60,831,480 [49.7%], Trump 59,556,646 [48.7%] [NBC]
Biden 64,668,371 [49.8%], Trump 63,253,073 [48.7%] [NBC]

NBC featured the same popular vote feature in its graphics--but on a much more frequent basis--in its continuing coverage 'three days later' (apparently Friday, Nov. 6, as it featured this press conference) (these figures are from across about an hour and a half of coverage):

Biden 73,920,086 [50.5%], Trump 69,802,341 [47.7%] [NBC]
Biden 73,925,498 [50.5%], Trump 69,804,128 [47.7%] [NBC]
Biden 73,933,205 [50.5%], Trump 69,811,875 [47.7%] [NBC]
Biden 73,944,419 [50.5%], Trump 69,831,326 [47.7%] [NBC]
Biden 73,959,694 [50.5%], Trump 69,846,397 [47.7%] [NBC]
Biden 73,963,998 [50.5%], Trump 69,856,738 [47.7%] [NBC]
Biden 73,977,930 [50.5%], Trump 69,873,504 [47.7%] [NBC]

---

For reference, it was around 11:30 PM Eastern on election night 2004 that Jeff Greenfield began drawing preliminary conclusions about the popular vote 'from the fact that every state has closed'.

I didn't watch any of these coverages live, but one thing that seems like it would have been shocking in 2020 is how quickly Trump relinquished the national popular vote lead. It didn't even last until 11:30 PM Eastern. In contrast, Romney's popular vote lead in 2012 lasted until just about half past midnight. Trump's lead in 2016 lasted until just about the beginning of the next waking day. (That said, McCain never led in the popular vote after the 11 PM poll closings, and wasn't leading even just before the 11 PM closings.) That obviously would have seemed an ill omen for the possibility of Trump winning the popular vote, but more than that, it would have seemed an ill omen for Trump winning the election, frankly. People seem to be beginning to think that the popular vote and the electoral vote are completely unmoored from one another, which, in theory, they could be. In practice, however, they aren't, even now. If Trump was going to be winning reelection, it seems he would have had to be keeping it at least as close in the popular vote as Romney kept it in 2012, at a bare minimum. (It may be worth noting that in every election the Republican has won this century--and in one that the Republican lost--the Republican was ahead in the national popular vote as of midnight in the East; when Jeff Greenfield observes that Al Gore has taken the national popular vote lead for the first time all night in 2000, a statement in the aston reads 'Official Count in Fla. Later Today', and a little later on, Judy Woodruff makes reference to 'this morning'.)

Another noteworthy thing is how constant the figures in 2004 are, and how well they matched up with the final results (50.73%-48.27%--or, to the ones place, 51%-48%). In contrast, overnight in 2008, the margin expanded from 3% to 5%. In 2012--after the West Coast had closed--the margin swung two points, from Romney +1 to Obama +1, which is still more of a swing than occurred overnight in 2004. In 2016, if they had been rounding to the ones place (as in 2004, 2008, and 2012), the margin would have swung overnight from Trump +2 (49%-47%) to even (48%-48%)--a two point swing. In 2020, it's hard to tell because after California closes, they don't give the percentages again until 12:47 AM. 

Furthermore, even the final overnight figures differed from the final results substantially in all these cases. In 2008, the final result was 52.93%-45.65% (or, rounded to the ones place, 53%-46%). In 2012, it was 51.06%-47.20% (or 51%-47%). In 2016, it was 48.18%-46.09% (or 48%-46%). And in 2020, it was 51.31%-46.86% (or 51%-47%). 

Perhaps what this demonstrates more than anything else is how much bluer California became in 2008 relative to even the Bush elections. In 2004, Kerry won California by just shy of ten full percentage points, 9.95%. Bush actually carried four of California's nine mega counties, and drew a > 100,000-vote margin out of one of California's counties (Orange). There were also twelve 'Bush '04 country counties' (casting at least 10,000 votes) in the state--that is, counties in which Bush got both over 60% of the vote and a higher vote share than Reagan in '84 or any nominee since. One of them--Kern County--cast well over 100,000 votes, and another, Tulare, cast just shy of 100,000 votes. California has never been a state, in recent times, where one metro area drags the state into the blue column (in the manner of Illinois or, to some degree, New York); Los Angeles County has never delivered California to the Democratic nominee despite its gargantuan size. (The last time the Democrat won California in one county was in 1948, and the county was Sacramento County.) However, if Los Angeles County cast no vote in 2004, California would have been somewhat close: 51.48%-47.18%. 

Now, after the 2016 election, Michael Barone wrote an argument that California's highly atypical voting relative to the rest of the nation, in conjunction with its great size, made the case for the Electoral College--and in doing so, he argued that the trend of California de-aligning from the rest of the United States was 'recent--and clear', and dated back to the 2000 election. Now, it should be noted that, even though the figures overnight in 2004 seem to match the final results well, the raw vote differential between Bush and Kerry overnight goes as high as 3,831,692 at one point (out of, obviously, a smaller number of total votes than the final total), whereas the final raw vote differential (out of a larger total number) settled down at 3,012,166. (And the percentages did poke up to 52%-48% at one point.) At one point in the evening, David Gergen, in underlining how strong a performance Bush had put in, referenced his popular vote lead as 'almost four million votes'. So it does seem as though California had some discernible impact on blunting the apparent power of Bush's popular vote margin as it appeared overnight. Going back to 2000, at a certain point on election night, Judy Woodruff and Jeff Greenfield point out that Al Gore has taken the lead in the national popular vote for the first time that night, due to votes from 'California and primarily LA County'.

However, things got much more dramatic from 2008 on. From 2008 on, no Republican has gotten more than 40% of the vote in California, and over half of the Republican vote in California has come from blue counties. In 2008, 2012, and 2016, there were no counties in the state in which the Republican was getting over 60% and a higher vote share than Reagan in '84 and every nominee since. (In 2020, finally, there were again--Tehama, which cast 28,731 votes, and Lassen, which cast 12,045.) Despite California having nine mega counties, Trump carried none of them in either 2016 or 2020. (McCain and Romney both carried one--Orange.) In 2012 (which featured the best Republican performance in California out of the last four elections), if Los Angeles County had cast no vote, Obama would have carried the state 56.3%-39.5%. 

It seemed that in 2020, a lot less attention was paid to the real-time national popular vote figures overnight than in 2004, 2008, 2012, or 2016. The national popular vote was displayed less regularly and/or without reference to percentages, and it seemed that it was discussed much less by panelists. This may have been a function of the fact that in 2008 and 2012, in particular, overnight speculation about the national popular vote based on real-time figures turned out to be significantly off. On election night 2008, for example, as the popular vote stood at Obama 51%, McCain 48%, Jeff Greenfield observed that 'for all that we thought that this might be a blowout, the popular vote is relatively close.' In 2012, Carl Cameron reported from the Romney campaign at about 11:30 PM Eastern that Romney's lead in the national popular vote was one reason for the Romney campaign's frustration at the early call of the national election; and just before half past midnight in the East, Megyn Kelly points out, in posing a question to Pat Caddell, that 'more [people] have voted for Romney tonight than have voted for Obama'.  

Already in 2016, at 12:23 AM in the East (as Trump held a 1.0% lead in the real-time popular vote), Chris Stirewalt raised the question of whether Democrats would accept the legitimacy of a Trump victory if he won the Electoral College without winning the popular vote. After Bret Baier, in response to this, suggests the panel look at the real-time popular vote figures, Joe Trippi says, 'I'll make a bold prediction here; as the West Coast comes in, [Hillary Clinton's] going to go surging by Trump in the popular vote.' In Barone's aforementioned article (published Dec. 5, 2016), he notes that California's slowness in counting its ballots was still 'obscur[ing]' the national popular vote outcome of the election (he felt comfortable enough to say that Hillary Clinton 'apparently won [the popular vote] by a slimmer margin than Samuel Tilden in 1876'). 

Compounding this, ahead of 2020, Republicans had largely resigned themselves to Trump losing the popular vote (but not to his losing the Electoral College). Unlike in 2004, Republicans 'downplay[ed] the symbolism of the popular vote', with Dan Eberhart arguing that it was 'irrelevant'. If a party openly embraces a strategy of winning the Electoral College despite expecting to lose the popular vote, there is that much less reason to heed the popular vote.

In the coverage of 2020, I didn't notice anyone mentioning the real-time popular vote until the next waking day, when Bret Baier made reference to the 2.0% margin (as it was at the time) at least twice in contrast to Fox News' last pre-election poll that had Biden leading in the national popular vote by 8% (with a margin of error of 3%). (It was also not until the next day that Fox News had the popular vote displayed on a really regular basis; and NBC also seemed to display the popular vote much more regularly three days after election night than on election night itself.) In this case, the popular vote had solidified enough to put Baier on reasonably safe ground in putting the polling industry in question on the basis of the real-time popular vote figures, although even here, Biden finally ended up with a 4.45% lead--still short of the 5% that would have been at the lower end of the range of 5%-11% allowed for by the margin of error, but not by that much.