A Plurality for "Did Not Vote"

I stared at this chart of U.S. presidential election results from 1976 to present for a long while today when I saw it on BlueSky:

Each year has the colors — and therefore what they represent — arranged from highest percentage to lowest, left to right. As you can see, "Did Not Vote" has consistently been the highest percentage, except in 2020 when Joe Biden just barely exceeded it. 

Political scientists, I'm sure, have many hypotheses about why 2020 — and to a lesser extent 2024 — had historically low percentages of non-voters. The increasing ease of vote-by-mail and extreme political polarization are two that come to mind. 

One interesting fluctuation in the usual rate of "Did Not Vote" was in 1992, which had the best turnout. That was when Ross Perot was running as a credible third-party candidate. Other than that, "Did Not Vote" was pretty stable between 1976 and 2000, running between 47 and 49% each year. (I don't know if the pre-1976 non-voting percentage was higher or lower, or if it fluctuated greatly.)

Then in 2004, after 9/11 and during the Iraq War, it suddenly dropped 8 points and has stayed relatively lower than the earlier consistent trend line. The question must be, why was that?

Some of it could be from relaxation of the rules restricting vote-by-mail, and the institution of voting completely by mail in some states (Oregon 1998, Washington 2009, Colorado 2013). If it's easier to vote, more people vote: It makes sense. 

The other thing that surprised me about the chart was how small the margins of popular vote victory were in some of the races, especially compared to "Did Not Vote." Obama got 33% of the vote in 2008, compared to McCain's 28%, while "Did Not Vote" got 39%. In the year of Hope and Change, 39% did not feel inspired to vote at all.

After Watergate, 47% did not feel inspired to vote and throw the rascal out who had pardoned Nixon. That's 20% more than who voted for Carter. The supposedly disastrous Dukakis only lost by 4 points, which I guess is a lot, but also sort of... not?

Lots of interesting things in this chart.