(This is off-topic for me, but I got a little obsessed with the topic. See UPDATE at the end.)
At first glance, it appears that the number of votes for the Democratic presidential candidate were very anomalous compared to the previous and succeeding elections.
I know that people have analyzed the 2020 results to death, but this interested me enough to look at it on my own.
First of all, the scale of the chart - while acceptable practice - is deceptive, since the Y axis starts at 50 million votes, exaggerating the effect. Here is the correct chart for the past five elections in proper perspective.
Obviously far more ballots were counted in 2020 than for any US election in history. The reasons given, according to ChatGPT:
Heightened Political Engagement: The political climate in the years leading up to the 2020 election was particularly charged, motivating more people to vote.Expanded Mail-In and Early Voting: Due to the COVID-19 pandemic, many states expanded access to mail-in and early voting to make it safer and more convenient for people to cast their ballots.Increased Voter Registration Efforts: There were significant efforts by various organizations to register new voters, particularly among younger demographics and minority groups.High Stakes: Many voters on both sides viewed the 2020 election as particularly crucial, leading to higher turnout.Intensive Get-Out-The-Vote Campaigns: Both major political parties and numerous advocacy groups invested heavily in efforts to mobilize voters.
Four of the five apply to 2024 as well, yet the number of votes cast went up by over 21 million in 2020 and went down by about 18 million this year. The main difference as far as I can see is that there was a huge increase in mail-in voting in 2020 due to COVID concerns - 65 million compared to 25 million in 2016.
This year there were more than 80 million early voters, but I cannot find a breakdown yet of mail in vs. early voting at official locations.
Looking further into the 2020 numbers, I looked at the differential of votes compared to 2016 by party. My theory is that in general, 2016 Trump voters remained Trump voters (49%) and 2016 Clinton voters remained Clinton voters (51%), within a couple of percentage points.
If we subtract the 2016 results from the 2020 results, the excess number of votes come out to 58% Democrat vs. 42% Republican (I'm not counting third parties).
That number is, to me, statistically suspicious. The additional votes should be roughly in line with the more general results (which totaled 52% Democrat to 48% Republican, including the excess votes.) Clearly the excess votes are way out if whack and not close to a representative sample compared to the rest of the voters in America.
Even more crazily, if you subtract the number of Democratic and Republican votes of 2024 from the historical highs of 2020, you see that the "missing" votes that weren't cast were 83% Democratic and 17% Republican. (In this case, however, there are stong indications that there was a lot of voters who "flipped" from D to R this year so this number is not as meaningful.)
Now, it is possible that there are non-suspicious reasons for the 2020 anomalies. Trump disparaged mail-in voting throughout his campaign so in general Republicans went to the polls while a far higher percentage of Democrats voted by mail.
Surveys afterwards showed that people were happy with the ease of mail-in voting. However, if it is so easy and convenient, why did we see the number of votes cast go down so dramatically this year?
Nevertheless, the claim that somehow the 2020 voting - specifically the excess voting - was manipulated would require extraordinary proof to me. My hypothesis to test that idea is to look at each individual state's excess voting. If the Democrats were concentrating on manipulating the votes somehow, it would be logical to expect that they would put far more effort into the battleground states and not bother much with states that were not up for grabs.
The eight states considered battleground states in 2020 were Arizona, Florida, Georgia, Michigan, North Carolina, Ohio, Pennsylvania and Wisconsin. When I look at the percentage difference between those eight states and the 2016 results, I do not see them as being towards the top in a list of anomalies. The top four gains for Democrats percentagewise compared to 2016 were in already blue states (New Hampshire, Maine, Maryland, Connecticut, all with above 24% difference in Democrat excess vote in 2020 compared to total in 2016.) The swing states are generally in the bottom half of that list.
Yet, when you compare some of the swing states percentage point difference between the excess 2020 votes and how they vote in 2016, some of them come out above the national average of 6.91 percentage points. They include Georgia (14), Michigan (9), North Carolina (7.4), Arizona (7.4) and Pennsylvania (7.) All of those besides North Carolina flipped from R to D.
Ohio, Arkansas, Nevada and New York's excess votes were fairly close to their 2016 votes.
Florida's excess votes were more Republican than in 2016 by 6.4 percentage points.
My analysis does not indicate that there was a concerted effort to manipulate the votes in the battleground states, at least no more than other states. Maybe many of the new 2020 Democrats, who were less motivated in previous years to vote, were unenthusiastic about Kamala Harris and went back to not voting.
But there are still many unanswered questions about the 2020 election and why it had so many more Democratic votes than the previous and successive election years, compared to Republicans.
UPDATE: Yair Rosenberg points out that the 2024 numbers (which I had verified at the time I first looked at this) have been significantly updated as new votes are counted. The number of votes in 2024 is not nearly as significantly lower as I thought - at the moment, less than a million fewer than 2020. So my comparison between 2020 and 2024 goes out the window, although why the excess votes in 2020 were so skewed towards Biden is still strange, to me at least.
Updated chart, including the section outlined in red shown in the original graphic for comparison:
